Claim frequency is a term often discussed in workers compensation (WC) circles. For many years, NCCI has reported on the long-term decline in claim frequency, citing automation, robotics, and continued advances in safety as contributing factors.
How Exactly Is Claim Frequency Defined?
Generally, the term refers to a ratio of claim counts to a selected exposure base during a specified period. However, there can be several variations:
- Claim counts can refer to:
- Indemnity claims
- Medical-only claims
- There are many choices for the selected base:
- Number of workers
- Claims can be aggregated by:
- Accident year
- Policy year
- Claims can be subdivided by:
- Industry Group
- Size of Loss
- Type of Injury
- Cause of Injury
- Claim counts may be as reported or developed to an estimated ultimate level
- Adjustments may be applied to the selected exposure base (e.g., wage adjustments)
WC stakeholders often ask how NCCI measures and monitors changes in frequency over time. In this article, we will examine the three frequency measures most often cited and published by NCCI along with the primary uses of each measure.
The example below shows the calculation of these three frequency measures for one year for a hypothetical employer and illustrates how these measures may vary considerably in magnitude. We define lost-time claims as those involving indemnity benefits.
Building upon the example above, the hypothetical example below compares our frequency measures for employers in two very diverse classes of business—a contractor and an office.
Below are key observations from the example above, which also hold true in WC experience:
- As expected, frequency per payroll and per worker are high for the contractor and low for the office, given the relative hazards inherent in these occupations.
- In contrast, frequency per premium is relatively low for the contractor. This result, which may seem counterintuitive, occurs because the contractor’s premium in the denominator is generally high relative to the other classes of business, due to higher-than-average frequency per worker and claim severity.
I. Frequency per Payroll
NCCI compiles this measure using lost-time claims, which typically account for more than 90% of total loss dollars. We can also compute this measure for medical-only claims and total claims.
To review changes in frequency per payroll over time, we adjust the payroll for each older year to the current year’s wage level by state.1 Without this adjustment, there would be downward pressure on frequency per payroll from one year to the next as payroll increases when wages go up.
Frequency per payroll provides a valuable “snapshot” comparison of frequencies by class of business for a given year, by state, or for all states combined.
- Class Code 8810 (clerical office) typically has very low frequency per payroll—usually about one-tenth that for all NCCI classes combined
- At the other extreme, Class Code 5551 (roofing) typically has very high frequency per payroll—usually more than five times that for all classes combined
This measure can also be used to determine year-to-year frequency changes for one class of business. However, when examining all classes combined, NCCI cautions that year-to-year changes in this measure may be very distorted due to shifts in the industry mix. For example, consider the two classes cited above. An increase in the share of Class Code 8810 or a decrease in the share of Class Code 5551 would put downward pressure on the frequency per payroll for all classes combined.
II. Frequency per Worker
NCCI compiles this measure separately for lost-time, medical-only, and total claims. We do not collect data on the number of workers. Therefore, we estimate the number of workers by dividing a state’s payroll by its average annual wage.2
Exhibit 12 of NCCI’s
Annual Statistical Bulletin displays frequency per 100K workers by NCCI injury type, by state, and countrywide. NCCI’s legislative pricing team uses the distribution of claims by NCCI injury type to estimate the impact of proposed and enacted changes in WC benefits by state.
NCCI cautions that, while state comparisons are possible, it is important to recognize that the mix of industry may have a significant influence on the magnitude of this measure. For example, of all US jurisdictions, the frequency per 100K workers is the lowest in the District of Columbia where office workers are predominant.
As was the case with frequency per payroll:
- Frequency per worker is relatively high for contracting workers and low for office workers
- Year-to-year changes in this measure can be distorted by shifts in industry mix
III. Frequency per Premium
NCCI compiles this measure using lost-time claims. Premiums for each year under review are adjusted to the current approved pure loss cost level in each state.3 Also, as with our frequency per payroll measure, we adjust for changes in wage level by state. We calculate this measure using Financial Call data as well as
Statistical Plan data.
Previous research published by NCCI identified the fact that frequency per premium varies far less by class of business than frequency per payroll or per worker.4 As a result, this measure is less impacted by shifts in class mix from one year to the next. For this reason, NCCI relies upon this measure for analyzing changes in frequency over time by state or countrywide.
- NCCI analyzes frequency per premium by policy year as part of the annual rate/loss cost filing review in each state. These frequencies are typically displayed in the NCCI rate/loss cost filing as well as in each state’s annual
State Advisory Forum presentation.5
State of the Line presentation includes the following frequency per premium information by accident year, which is updated annually:6
- Changes in frequency per premium for all NCCI jurisdictions combined
- Average annual change in frequency per premium by state over the most recent five years
The analyses above are based on our Financial Call data. Alternatively, using
Statistical Plan data allows us to slice frequency per premium by various claim and employer characteristics. This enables us to determine if the observed overall frequency changes are widespread, confined to certain segments, or differ considerably by segment.
An External Frequency Measure
The Bureau of Labor Statistics (BLS)7 publishes a frequency measure defined as the number of injury and illness cases involving days away from work per 10K full-time equivalent workers. The information is collected annually in its Survey of Occupational Injuries and Illnesses. NCCI closely monitors this measure for consistency with our data.
NCCI will continue to monitor trends in frequency and share our findings.
This article is provided solely as a reference tool to be used for informational purposes only. The information in this article shall not be construed or interpreted as providing legal or any other advice. Use of this article for any purpose other than as set forth herein is strictly prohibited.