Copyright © 2004 Business Insurance

"Avoid Pitfalls in Benchmarking for Savings" March 1, 2004

by George G. Pallis

Many organizations are making greater use of benchmarking in their risk management efforts. Gaining the most from these activities requires a clear understanding of the objectives and a careful consideration of how a benchmarking analysis should be conducted.

Benchmarking helps firms evaluate program costs, market capacity, retentions, deductibles, attachment points, limits, sublimits and coinsurance. In addition, it aids in the examination of coverage scope and exclusions, insurer preferences and service capabilities, risk-financing options, and claim and litigation trends.

With underwriters requiring detailed information on clients' losses, exposures and relevant loss control activities, the effective use of benchmarking to identify and control cost drivers can help in negotiating better quotes, terms and conditions. Additionally, a detailed benchmarking report might improve insurers' perception of a firm's risk profile.

When incorrectly performed, though, benchmarking can lead to the wrong conclusions or prevent an organization from obtaining the best results.Here are seven common mistakes that can produce those outcomes:

-  Inadequate sample. Many benchmarking data sources can provide breakouts for multiple revenue groupings, industries, locations and other categories. The sample size must be adequate for each of the breakouts to be statistically significant, though. Consider workers compensation, for which rules and regulations vary by state. Relevant data for benchmarking typically must be analyzed by jurisdiction as well as by employee occupation. This requires several segmentations of the data. In addition, workers comp benchmarking efforts typically call for examining frequency and severity, specific business units or operations and individual exposures rather than aggregated risk. Therefore, it is important to know the size of the sample.

-  Not finding the right peers. A peer might be an organization with the same scope of operations within a firm's industry group; an employee base in the same occupation, regardless of industry; or claims data with the same loss characteristics.

To identify peers, start with your organization's benchmarking objectives. It's not unusual to select different peers for different benchmarking objectives. In many cases, a firm in the same revenue class might have a different scope of operations. To benchmark auto risk, fleet size often is more meaningful. And payroll may be a better yardstick for workers comp. To illustrate this further, a firm that analyzed its workers comp costs measured itself against one peer and saw no need for change. Yet, because it manufactured two vastly different products, it conducted a second analysis against a separate peer, which helped it pinpoint significant opportunities for savings.

-  Limited access to data. Data for use in benchmarking is available from a range of sources. For internal comparisons, firms collect their own data.

To benchmark against peers, they often must obtain data from third parties, including insurers, brokers, trade groups and management consultants. Yet these sources may not provide the same access to data. Accessibility to raw data from the central database of a broker or consultant often enables companies to obtain better and more meaningful custom reports on a timely basis. These sources may be especially valuable for loss and insurance program structure information, because they may include information aggregated from multiple insurers.

-  Poor data quality. While self-reported data accumulated by third-party sources may be of some value, aggregated data from actual bound programs gathered by insurance brokers and consultants from their client bases usually provides greater assurance of quality and accuracy.

-  Outdated data. Significant changes in the insurance marketplace during the past several years may make historical comparisons that draw on data that is two or three years old less valuable. Consequently, knowing the timeliness of the data is extremely important. In addition, the ability to examine data by time period can be very useful.

-  Overlooking potentially meaningful trend data. A key aspect of many benchmarking projects involves gathering baseline data relative to an organization's risks and costs. Although the data could encompass a single period, such as one year, compiling data in successive years often yields a clearer picture of a firm's risk profile and the effectiveness of risk control, productivity and process improvement initiatives over time.

-  Not using the right measurements. When evaluating benchmarking data or reports from third parties, make sure you use meaningful statistical measurements. Some benchmarking reports cite only minimum, maximum and average (or mean) values, which can be misleading. The median, or median range, can reduce the distortion in means caused by outliers. In addition, benchmarking insurance purchasing trends and program structure often calls for going beyond the categories that may appear on a survey form. Underlying factors in these areas should be explained in the accompanying text or as footnotes to the charts.

Properly used, benchmarking provides firms with a multitude of benefits. It helps evaluate the effectiveness of initiatives designed to generate savings or improve performance. In this respect, it should not be a single exercise but a tool to yield continual improvement.

George G. Pallis is a senior vp at Marsh Inc. in New York.

 

© Copyright Business Insurance 2004