Introduction

The Joint Committee to Oversee the Professional Salary Evaluation System was formed in late 1996 to oversee the current project to collect data from UMPSA employees about their jobs; apply current market data to benchmarked jobs; evaluate the job data and create an evaluated model hierarchy of jobs using WYCOMP Evaluation Software (from Watson Wyatt Consultants); confirm the evaluation model is based on the current market value of the jobs; and to produce documentation of this effort. The July, 1995 to June 30, 1997 UMPSA Bargaining Unit Agreement, Article 17, Section D. 1., set forth the terms by which the Oversight Committee came into existence, and authorized the project to develop a Professional Job Evaluation System (using Watson Wyatt Worldwide software referred to as WYCOMP) to continue.

The Joint Committee membership is Tracy Bigney, Casandra Fitzherbert, Bruce Littlefleld, Lynda Kinley, Catherine Pease, Mary Skaggs, Elise Scala, Sheri Stevens.

The Project Methodology

Two professional fixed length employees [Edward Flotten and Gina Toman] were hired to assist Larry Pile, Compensation Specialist with the System Office of Human Resources, in completing the project. These three employees performed the evaluations for all jobs and set up/maintained overall project continuity. Utilizing a project timeline, the questionnaire for collecting employee job data was edited slightly to clarify some questions - but no substantive edits were made to the questionnaire. Employee clarification meetings and questionnaire distribution, Watson Wyatt training for WYCOMP software use, and follow-up communication procedures were established during the project. Data input procedures were created and the selection of benchmark jobs, with assistance from Watson Wyatt, was completed by the evaluators. Evaluators reviewed and added material as appropriate to the uniform written guidelines for evaluating jobs using WYCOMP software. An evaluation cross checking procedure of completed jobs was instituted to ensure uniform evaluation methodology.

After the benchmark selection process, two evaluators used the Watson Wyatt market salary survey library in Boston for collection of current market data for over 100 of the benchmark jobs. The evaluation group used a uniform methodology in applying the appropriate market data to the appropriate benchmark job. Market data was "aged" as appropriate to reflect market survey data and the benchmark employees' full-time equivalent salary as of April, 1997.

The evaluation process involved using the WYCOMP software to derive a point value for each position rather than a market currency value for the job. The process can be summarized as follows:

(1) A job evaluation questionnaire was completed by the employee and supervisor. Any differing interpretations of questions and job content were resolved at this step.

(2) The questionnaire data was entered into the WYCOMP program database and the data entry was verified.

(3) The responses were reviewed using WYCOMP software by the evaluation team members. The responses were reviewed for internal consistency and were compared to similar jobs in the Professional Unit, helping to maintain equity with comparable jobs. The result of this step is a point value and resulting salary grade for the position.

(4) The point value generated by the job evaluation process placed the position in one of several salary grades.

The "job evaluation" process within WYCOMP is based on a statistical model which uses the questionnaire responses to predict a "point value" for each position. An equation assigns a weight to each individual questionnaire response or a combined group of questionnaire responses. The sum of the weighted responses used in the WYCOMP software had been determined statistically by Watson Wyatt in our 1990-91 original evaluation model. The weights reflect no individual judgments or biases. The weights were those which most accurately predicted the market hierarchy as of 1990-91 - the order in which positions were valued in the hiring market.

Watson Wyatt developed the salary structure (salary grades) by using data from the generated point values and the assigned market data of benchmark positions. The salary structure is made up of several grades with an associated minimum, midpoint, and maximum salary assigned to each grade.

The project evaluation phase and the resulting point value job hierarchy with the wage structure was completed in December, 1997.

Summary/Observations

The evaluation team sent out 979 questionnaires to eligible UMPSA employees. 677 were returned and 593 were evaluated and placed in the current model with appropriate wage grades and salary information. 84 questionnaires were not evaluated because the employee was not in a unit position, the questionnaires were sent back to the campus for further work, the evaluators could not resolve differences between employee and supervisory responses, the job no longer existed, the job duties had changed dramatically, or follow-up communication was not returned.

The salary structure developed by Watson Wyatt (based upon the listing of benchmark jobs in the model, benchmark job predicted values, and the market data for each of these benchmark jobs) was seven wage grades with a 10% midpoint spread (the percent difference between two consecutive midpoints) and 50% range spread. Watson Wyatt advised the committee that the newly created salary structure did not reflect the evaluated points as being a good predictor of the market. A Watson Wyatt regression of the new wage grade assignments against the available estimated market values showed a relation of R squared .291 (the higher the statistical R squared value the closer the relationship and explanatory rationale is for comparing two sets of data). Watson Wyatt advised the very weak relationship of point value grade to market data is occurring because our factor model weights (used in the WYCOMP evaluation calculation) are not reflecting the true representative market values today. The UMS WYCOMP job evaluation system does not reflect the current market.

In summary, Watson Wyatt advised the committee that the WYCOMP model predicted point values do not have a close relationship to market values. Our factor weights have not maintained pace with the compensable factors now valued by the market.

The Committee recognizes that a somewhat less than satisfactory business relationship with Watson Wyatt also was noted during the project. This may have resulted in part from the length of time required to complete the project. The project lost the principal Watson Wyatt consultant, Rachel Brown, who worked with the project during 1989-97 early in 1997. The project experienced numerous programming bugs and ultimately lost technical support for the WYCOMP program in July of 1997. The WYCOMP software for the project was no longer being maintained or sold by Watson Wyatt and had been replaced by the windows based REWARDS system. In essence the UMS project software was considered outdated/obsolete by Watson Wyatt. The cost to update to the Watson Wyatt REWARDS software would be in the $50-70K range and would involve virtually restarting our project.

The Oversight Committee notes that the basic data collection instrument, WYCOMP Evaluation Program, and evaluation methodology used in this 1996-97 model remains substantively the same as that used in the 1990-91 evaluation model submitted with the February, 1996, Final Report of the Joint Committee to Design a Professional Salarv System. Market data collection for the current project was performed by the evaluation team rather than by Watson Wyatt consultants and job evaluation was much more uniform with only 3 evaluators actually reviewing jobs.

The attached report entitled 1996-97 UMSPA Evaluation Project Summary summarizes the project events and evaluators observations, lists evaluated jobs and benchmark jobs, lists the final salary structure, market, job family, and gender data , and summarizes the salary/benefit costing to raise represented professional employees to the minimum of the salary range and the midpoint of the salary range. We have also attached letters from Watson Wyatt dated November 7, 1997, and December 2, 1997, enumerating the evaluated points versus market pay weakness discussed earlier in this report.

The Oversight Committee notes that the estimate to bring current unit members to the minimum of the salary grade predicted by the WYCOMP system would be approximately $1.39 million. The approximate cost to bring current unit members to the midpoint of the assigned grade would be $6.48 million. These figures are for salary and benefit estimates only. The figures do not include substantive administrative costs for implementation.

The Oversight Committee, recognizing that Watson Wyatt advised that the current internal job evaluation system has a very poor relationship between predicted job value and current market data, requested contractually that Watson Wyatt sell the source code for the WYCOMP software to the University. At the suggestion of Watson Wyatt the Committee reviewed the software documentation. Based on this review the Committee determined that purchase of the source code was not cost effective.

On behalf of the Committee Bruce Littlefield conducted multiple regression and neural network analyses of the questionnaire data to determine whether revised weightings of the data would result in a model which better reflects market values. The Committee reviewed the correlation coefficients and residuals from these models and found that the results did not show sufficient promise to warrant further investigation.

Conclusion

The tasks which the University was charged to complete in this project have been accomplished. The Oversight Committee finds that the WYCOMP system does not adequately reflect values of the current market in predicting job grades. The external consultant, Watson Wyatt, concurs in this conclusion.

The Committee pursued methods of revising WYCOMP to improve the relationship to market and did not find a significant improvement. Based on these findings the Oversight Committee recommends that the WYCOMP system not be implemented. However, the Committee recognizes that there continues to be a need to improve the way in which salaries for unit positions are determined.

This report is submitted in accordance with Article 17, Section D. 1., of the 1997-98 UMPSA Unit Collective Bargaining Agreement. The Agreement provides that "either party may serve upon the other a notice of at least thirty (30) days no earlier than January 1, 1998, advising that they desire to confer and negotiate with regard to the proposed salary system."