HD6 - Technical Report: Gender Pay Equity in the Virginia State Workforce


Executive Summary:
The Department of Personnel and Training (DPT), using October 1998 data, replicated the statistical analysis that the Joint Legislative Audit and Review Commission (JLARC) conducted in 1997. The 1998 data set included information on 65,816 employees who were subject to the provisions of the Virginia Personnel Act. Of these, 33,748 (51.3%) were female and 32,068 (48.7%) were male.

The data was grouped into 6,086 agency-class combinations and salary averages were computed for males and females in each agency-class. DPT applied screens that JLARC developed to the agency-class summaries to identify possible cases of gender-based discrimination. Sixty-one agency-class combinations (1.0%) passed through the screens and, thus, were identified for further examination. Of the 61, thirty-three involved males being paid more than females and 28 involved females being paid more than males.

The percentage of cases where the screens did not explain salary differentials between genders was small (1.0%). Also, there was balance between the number of cases where males were paid more (33, or 0.54%) and where females were paid more (28, or 0.46%). These findings do not indicate that the State's compensation program violates the principle of pay equity for similar work.

The State system is highly structured and controlled by policies. Thus, there are few opportunities for gender-based discrimination. Starting pay, competitive offers, reallocation increases, and performance increases were identified as the only pay decisions where agencies' managers have the discretion to make decisions affecting the relationship of males' and females' salaries.

The 61 agency-class observations that passed the JLARC screens included 188 employees in 41 agencies. DPT contacted these agencies for qualitative information to explain why the males were paid more than the females, or vice versa. Thirty-nine agencies responded to the survey. The reasons for pay disparity fell into 8 categories: starting pay, prior experience, performance increases, competitive offers, northern Virginia differentials, length of service, administrative error, and transactions sequence. No indication was found that employees' genders formed the basis for pay disparity in any agency.

The final JLARC recommendation was for DPT to review and update its job classification system. A major effort, known as the Class Specification/Specification Update (CWSU) program attempted to do this in the latter 1980's and early 1990's. The program was since dropped due to DPT staff reductions. Maintaining specifications for agency-unique classes has been decentralized to the agencies that use them.

Current efforts may result in an updated classification and compensation program and may improve pay equity. Included are the work of the Commission on Reform of the Classified Compensation Plan and compensation pilot programs in place in selected agencies. Also, DPT has proposed a simplified classification and compensation approach for information technology positions; this approach could be expanded to other occupational areas. Finally, DPT has drafted an in-range salary adjustment policy, which would allow adjustments of employees' salaries for reasons including improved equity among employees.

JLARC asked that DPT assess the placement of job classes in grades 7 through 11, and whether the implicit tradeoffs between different job requirements, such as education and working conditions, are appropriate. The assignment of salary grades is based a complex combination of factors. The first factor used in assigning salary grades to job classes is internal alignment. Each job class is compared with other classes in the same general occupational area. Seven factors are used to evaluate the job classes: Complexity of Work; Supervision Given; Supervision Received; Scope; Impact of Actions; Personal Contacts; and Knowledge, Skills and Abilities.

DPT does not utilize comparisons of the classification factors of dissimilar jobs in assigning salary grades, and it does not support the use of this methodology for evaluating the appropriateness of ranges assigned to male-dominated job classes vis-à-vis female-dominated classes. It is simply too subjective to compare dissimilar jobs, where the grade assignment of one class may be based largely on one classification factor while the grade of another may be based on an entirely different factor.

There are other indicators, rather than job evaluation factors, that can be used to evaluate whether the grade assignments have an adverse effect on either gender. Market data from southeastern states and from Virginia private employers was used as the primary indicator of the appropriateness of the salary ranges of male-dominated and female-dominated classes in grades 7 through 11.

Overall, the survey of southeastern states indicated that employees in male dominated classes, on average, were paid more than were employees in female-dominated classes relative to the other states. The simple average deviation was 0.53% higher for male- dominated classes, and the weighted average deviation was 0.96% higher. However, neither group's deviations were as high as the deviations of the non-dominated group.

Similarly, the state survey of private industry found that the male dominated classes were more competitive than the female-dominated classes. In this case, the male-dominated classes had smaller negative deviations. The simple average deviation was 4.99% lower for male-dominated classes, and the weighted average deviation was 1.65% lower.

Turnover rates did not explain the differences among the salary range deviations of the three groups of classes. The 19 female-dominated classes had the lowest salary ranges relative to the other states, but the highest average turnover rates, with a 13.55% average weighted by the number of employees in each job class and a 13.36% simple average.

As the classification and compensation program is updated, DPT should ensure that any differences in salary ranges between male-dominated classes and female-dominated classes are supported by job evaluation criteria (within occupational areas) and by market and staffing data.