Process Changes in Unemployment Insurance Claims Statistical Reporting

Each month in Wyoming Labor Force Trends, Wyoming Unemployment Insurance Statistics (Initial and Continued) are published for initial, additional and continued unemployment claims. These charts and tables show a breakdown of claims by county and industry for the current month as compared to the previous month and the current month one year ago. After a claim has been filed it is placed in a database for use in analyzing unemployment tendencies and economic trends of localities and industries. Claims are grouped by week regardless of when one month ends and the next month begins. This grouping procedure can cause a misrepresentation when comparing the volume of claims between subsequent months, or comparing the same month from year to year.

Problem: No Standardization

Once claims have been entered into a database the analysis of data can begin. Using a calendar, the number of "complete weeks" for the current month are counted. A "complete week" is defined as one that has four or more days in it. For example: the week of August 1, 1996 has four days in July and three days in August. This week would be identified with July because July has four days in the week. Thus, August 1996 would have four complete weeks, while July 1996 would have five. Claims associated with the weeks are compiled and identified with each respective month. This process is repeated for the current month one year ago. These data are then aggregated and placed into the tables and charts shown in Trends (see Initial and Continued Claims).

Because we have not standardized months with respect to weeks, there has been a problem with the usual approach to organizing claims data for economic analysis. The charts and tables compared the current month to the previous month and also to the current month one year ago. But the analysis has not accounted for the fact that the months compared may have different numbers of complete weeks. For example, in the October 1996 issue of Trends a comparison is made between August 1996, July 1996 and August 1995 (see Table 1 for a partial reproduction of this data). However, August 1996 had only four complete weeks while both July 1996 and August 1995 had five complete weeks. In a comparison of continued claims, July 1996 and August 1995 had considerably more weeks claimed simply because of the extra week and not because of economic conditions. So, while it may be true that fewer claims were filed in August 1996, the volume of the claims as shown in Table 1 may be inflated simply by virtue of the fact that July 1996 was longer than August 1996 by one week.

Solution: Normalization

To solve the problem of unstandardized months, we have to change the way we manage and process the claims data. Once the weekly claims have been aggregated into the total number of claims for the month, we divide it by the number of complete weeks in the month and then multiply the result by 4.33 (52 weeks/12 months=the average number of weeks in a month). The formula is:

The result is claims data for each month that has been adjusted to compensate for an unequal number of weeks per month. The months have been normalized according to the number of weeks (see Table 2).

Analysis

The change in data processing has a major impact on the distribution of the data over time. After normalizing the data, in the category total weeks claimed for August 1996, the number of weeks claimed increased by 838, while July 1996 and August 1995 decreased by 1,954 and 1,739 respectively (see Table 3). Overall, August 1996 weeks claimed increased by 8.3 percent whereas both July 1996 and August 1995 weeks claimed decreased by 13.4 percent (see Figure 1).

Before the change in processing, there was a 30.4 percent decrease in weeks claimed from July 1996 to August 1996. After the change in processing to normalizing, the decrease in weeks claimed was only 13.0 percent. The 13.0 percent decrease in weeks claimed with the normalization process is consistent with the decline in weeks claimed when analyzing July through September 1996 data with the new normalization method. Weeks claimed decreased 10.2 percent from June 1996 to July 1996 and decreased 17.4 percent from August 1996 to September 1996 (see Table 4). The normalized process shows a steady decrease in weeks claimed compared to the erratic behavior using the unstandardized method (see Figure 2).

Conclusion

Normalizing the data enables the comparison over time to reflect claims activity induced by changes in the economy rather than by changes in the units of time over which the measurement takes place. By removing the impact that the number of complete weeks has on the data, there is no longer an inflated number of claims for the months with more complete weeks or an unnaturally low number of claims for the months with less complete weeks. All months will have the same amount of weeks so that the comparisons can be taken at face value.

Mary Beth O'Loughlin is a Statistician specializing in Mass Layoff Statistics (MLS) and Unemployment Insurance (UI) with Research & Planning.

Go Back To:

Table of Contents

Labor Market Information

Employment Resources

Send Us Mail