Back to Top

Level of Comparability Payments for January 2004 and Other Matters Pertaining to the Locality Pay Program

Thursday, October 21, 2004
From: 
Federal Salary Council 1900 E Street NW Washington, DC 20415-8200
Subject: 
Level of Comparability Payments for January 2004 and Other Matters Pertaining to the Locality Pay Program

As authorized by the Federal Employees Pay Comparability Act of 1990 (FEPCA), we present our recommendations for the establishment or modification of pay localities, the coverage of salary surveys conducted by the Bureau of Labor Statistics (BLS) for use in the locality pay program, the process of comparing General Schedule (GS) pay to non-Federal pay, and the level of comparability payments for January 2004.

Bureau of Labor Statistics Surveys and Pay Gap Methodology

Three of the five improvements designed for the National Compensation Survey (NCS) program are included in the surveys we reviewed thisyear.  These are:

  • Excluding data for non-Federal jobs that would be graded above GS-15 in Government,
  • Use of a crosswalk between Federal jobs and the Standard Occupational Classification System (SOC) with updated GS employment weights (March 2001), and
  • Introduction of an econometric model to estimate salaries for jobs that were not found in random samples of non-Federal jobs.

The Council is pleased to note that improved NCS surveys consistently yield results by grade that track fairly well with those based on the former Occupational Compensation Survey Program (OCSP).  However, two of the improvements are not yet included in the NCS surveys.  These are:

  • Better methods for grading supervisory jobs, and
  • Introduction of a simplified 4-factor grade leveling system with occupational specific guidance for 20 job families.

BLS is conducting final tests of the improvement in grading supervisory jobs in October 2002 and plans to incorporate the improvement in surveys conducted late next year.  Introduction of the new four-factor grading system will also begin next year, but will require 5 years to fully implement because of BLS' sampling cycle.  BLS conducts full-scale survey interviews only when it first adds an establishment to the survey sample and replaces only 1/5 of the survey sample each year.

While the NCS program results track the OCSP program results fairly well by grade, NCS averages are generally several percentage points below the OCSP results for grades GS-3 thru 7 and for grades above GS-11. While the overall average pay gaps are only 4 percentage points apart, the results vary significantly for a number of locality pay areas.  There are many factors that could cause pay measures under NCS to be different from those under OCSP.  Office of Personnel Management (OPM) staff identified a number of possible reasons for this outcome, including the following:

  • OCSP data are out-of-date, and the nationwide rate of change measures (i.e., Employment Cost Index) used to age the data probably overestimate or underestimate pay on a locality or occupational basis.
  • OCSP used a fixed job list that may have been biased toward higher-paying jobs.
  • Certain key OCSP results are based on small samples and may overstate results.
  • Test surveys indicated problems in assigning the appropriate grade under the NCS program.
  • NCS random samples may miss key high-paying jobs that are not common in non-Federal establishments.
  • Between 30 and 80 percent (varies by area) of the weighted data in NCS are modeled, and a review of the BLS model indicated that it tends to underestimate pay for high-paying jobs.  (In comparison, we had actual survey data for about 70 percent of the OCSP jobs and modeled about 30 percent, but OCSP survey jobs directly represented only about 30 percent of the Federal workforce.  It could be argued that actual data under OCSP truly represented only about 21 percent of the Federal workforce—about the same as actual data under NCS.  The OCSP model also tended to underestimate pay for certain high-paying jobs.)
  • Job definitions under OCSP were written to match specific Federal jobs, while the SOC crosswalk used in NCS has some more generic matches.

The Council wishes to thank BLS for its hard work in designing and introducing the improvements.  However, because the NCS improvements are not fully implemented and because there are substantial differences in survey results among areas, we recommend that the NCS results be phased in by averaging the pay gap for each area under NCS with the corresponding OCSP pay gap. Attachment 1 shows the pay gaps for March 2002 for both the OCSP surveys and the NCS surveys and the averages of the OCSP and NCS pay gaps based upon the Council's recommended approach.

Locality Rates for 2004

Based on calculations provided by OPM staff in averaging the two sets of pay gaps, the overall gap between base General Schedule average salaries (excluding any add-ons such as special rates and existing locality payments) and non-Federal average salaries surveyed by BLS was 32.02 percent as of March 2002.  The amount needed to reduce the pay disparity to 5 percent (the target gap) averages 25.73 percent.

We calculate the pay gaps excluding existing locality payments because locality pay is paid on top of the base General Schedule rates.  The overall average pay gap in 2002, including the current average locality rate of 10.93 percent, is about 19.0 percent.  The calculation is (132.02/110.93-1) X 100.

Under 5 U.S.C. 5304(a)(3)(I), the percentage of comparability payments due in January 2002 and any year thereafter may not be less than the full amount of the target gap.  Therefore, we recommend overall average locality rates of 25.73 percent for 2004.  We cannot calculate the percentage increase over the average of the rates authorized for 2003 at this time because the 2003 rates have not yet been set.  However, the Council points out that these rates are 1.86 points below the 27.59 percent average rate recommended by the Council for 2003 based on the OCSP surveys.  The proposed comparability payments for 2004 for each recommended pay area are also shown in Attachment 1.

These locality rates would be in addition to the increase in General Schedule base rates under 5 U.S.C. 5303(a).  This provision calls for increases in basic pay equal to one-half of one percentage point less than the percentage by which the Employment Cost Index (ECI), wages and salaries, private industry workers, increased between September 2001 and September 2002.  The ECI for September 2002 will not be published until October 31, so the amount is not known at this time.

Locations with Pay Gaps Below the Rest of U.S. (RUS) Pay Area

We previously recommended that locations with little data available in BLS surveys and pay gaps 2/10 of a percentage point or more below the RUS pay area or below the RUS pay area for three surveys be dropped from the BLS surveys, with the resources redirected to survey new locations. Under OCSP, the pay gaps in Huntsville, Indianapolis, and Kansas City are below that for the RUS locality pay area this year, and have been for several years. Under NCS, pay gaps for St. Louis, Orlando, and Kansas City are well below that for RUS, and pay gaps for Milwaukee, Dayton, Richmond, and Columbus are slightly below RUS.  With averaged results, Indianapolis, Kansas City, Orlando, and St. Louis are below RUS.

The Council's previous recommendation to drop locations more than 2/10 of a percentage point below the RUS pay area was intended to reallocate survey resources to survey a new city where the pay gap might be above the RUS pay area gap.  Since not all of the improvements have been completed in NCS and because the list of cities below RUS varies depending on whether we use OCSP or NCS data, we recommend that none of these locations be dropped as separate pay areas at this time.  The Council plans to consider whether some of these areas should be dropped and the resources used to increase the sample in other locations or used to add a few new locality pay surveys.  This review will be a part of our general review of other major aspects of the locality pay program scheduled for next year.  In this regard, BLS will begin planning for its geographic sample redesign in relation to new metropolitan area definitions in the fall 2003/spring 2004 time period.  This would be the optimal time for the Council to recommend and the President's Pay Agent to make changes in the geographic scope of locality pay surveys.  In the meantime, we recommend that the pay gaps in locations below RUS using the average of NCS/OCSP be combined with that for RUS in a cost-neutral fashion for the 2004 locality payments, as shown on page 2 of Attachment 1.

Areas of Application

Several areas no longer meet the area-of-application criterion of 2,000 GS employees for counties (or 1,000 GS employees for installations that cross locality pay area boundaries).  These are: New London County,CT; Santa Barbara County, CA; and Edwards Air Force Base, CA--all of which the Council previously recommended and the Pay Agent approved as areas of application to existing locality pay areas.

We continue to believe that it would not be prudent management or sound compensation policy to exclude New London, Santa Barbara, or Edwards AFB from their respective locality pay areas at this time.  In the past, we have recommended that once an area of application has been approved, it should not be removed for the duration of FEPCA's 9-year phase-in for the locality pay program (1994 through 2002).  While that time-line was up in 2002, locality pay is far from phased in.  In fact, an average of only about 42.3 percent of the target pay gaps was closed in 2002. Also, new census data will soon be available that will affect locality pay area definitions in the future.  Finally, the Office of Management and Budget (OMB) and the Census Bureau are revamping how metropolitan areas are defined.  These changes may also have an impact on pay area definitions.  We plan to conduct a major review of all aspects of the locality pay program next year when this new information is available.

Therefore, we believe New London, Santa Barbara, and Edwards AFB should continue to be included in their respective locality pay areas at the very least until we complete our review of the program next year.  Thus, New London County would remain in the Hartford locality pay area; Santa Barbara County and Edwards Air Force Base would remain in the Los Angeles locality pay area; St. Marys County would remain in the Washington, DC, locality pay area; Rhode Island and all of Bristol County would remain in the Boston locality pay area; and Monterey County would remain in the San Francisco locality pay area.

Requests for New Areas/Areas of Application

OPM staff were contacted by employees from over 40 areas since last year.  These contacts ranged from employees in Austin, TX, seeking to become a separate pay area to employees in Steamboat Springs, CO, concerned about high living costs.  The Council received four petitions this year.  Much of our August 15 and October 1 meetings were devoted to testimony about establishing four additional areas of application.  These were:

  • A proposal to include 21 townships in western Massachusetts in the Hartford locality pay area;
  • A proposal to include Barnstable County, MA, in the Boston locality pay area;
  • A proposal to include Monroe County, FL, in the Miami locality pay area; and
  • A proposal to include Larimer County, CO, in the Denver locality pay area.

This is not the first time employees in these areas have contacted the Council.  All have sent in proposals to previous Councils.  In 2000 and 2001, the Council concluded that it would recommend no changes in locality pay area definitions until it had an opportunity to review new commuting pattern and population data from the 2000 census and the new metropolitan area definitions to be produced by the Census Bureau and OMB in 2003.  However, we believe it was critical that the five new members of this Council hear first-hand about theses areas.

The Council's Working Group reviewed a sizeable amount of data provided by OPM staff on these areas.  Data included information relating to our current criteria, such as the number of GS employees, population density and commuting patterns, and other information, including quit and accession rates, losses to other areas, use of special rates, and mileage to the pay area.  While none of the areas seeking to be new areas of application pass the existing criteria, the Working Group recommended that the Council review each case separately on its own merits. 

In the case of Barnstable County, MA, it was noted that it is the only county on the eastern seaboard from southern Maine to Delaware that is not in a separate locality pay area.  It is also cut off from the remainder of the country by the Boston locality pay area and passes all our criteria except that it does not have 2,000 GS employees.  While we do not believe this is the time to make wholesale changes in locality area boundaries, the Council concluded that something should be done for Barnstable County.  Therefore, the Council as a whole voted at its October 1 meeting to recommend that the Pay Agent make an exception and include Barnstable County, MA, in the Boston locality pay area in 2004 in order to remedy what we consider to be an egregious situation.  The Vice Chair of the Council dissented from this recommendation.

Locality Pay Areas for 2004

We recommend continuation of the 32 existing locality payareas, as follows: 

  1. Atlanta, GA
  2. Boston, MA
  3. Chicago, IL
  4. Cincinnati, OH
  5. Cleveland, OH
  6. Columbus, OH
  7. Dallas, TX
  8. Dayton, OH
  9. Denver, CO
  10. Detroit, MI
  11. Hartford, CT
  12. Houston, TX
  13. Huntsville, AL
  14. Indianapolis, IN
  15. Kansas City, MO
  16. Los Angeles, CA
  17. Miami, FL
  18. Milwaukee, WI
  19. Minneapolis, MN
  20. New York, NY
  21. Orlando, FL
  22. Philadelphia, PA
  23. Pittsburgh, PA
  24. Portland, OR
  25. Richmond, VA
  26. Sacramento, CA
  27. St. Louis, MO
  28. San Diego, CA
  29. San Francisco, CA
  30. Seattle, WA
  31. Washington, DC; and
  32. Rest of U.S.--consisting of those portions of the continental United States not located within another locality pay area.

Status of Improving Future Surveys

Over the last several years, the Council has reviewed and monitored progress by Pay Agent and BLS staff in designing and implementing improvements in the NCS program.  The Council had recommended in its October 22, 1999, letter to the Pay Agent that these improvements should be made.  The Pay Agent submitted a report to Congress on May 15, 2001, on the status of these improvements.  The improvements cover:

1) Assigning GS grades to randomly selected survey jobs.

Progress:  OPM has completed development of a four-factor evaluation system for use in the surveys, and BLS has successfully used the new approach in field tests.  BLS will begin to phase the new approach into BLS surveys in December 2003.  This improvement will take the longest to implement.

2) Assigning GS grades to randomly selected survey jobs with supervisory duties.

Progress:  BLS and Pay Agent staffs have designed a new approach based on grading the highest level of work supervised and adding one, two, or three grades based on the level of supervision.  Final tests of the new approach are to be conducted this fall, and BLS hopes to be ready to implement the methods in surveys conducted in December 2003.

3) Other problems associated with random selection of survey jobs.

Progress:  BLS has designed and implemented an econometric model to estimate salaries for jobs not randomly selected in the surveys.  The model is derived from survey data and estimates pay for missing jobs as a function of location, occupation, and grade level.  The first production model, derived this year, explains 81 percent of variations in pay and has a mean absolute error of 17 percent. 

4) Matching Federal and non-Federal jobs.

Progress:  OPM formed an interagency working group that developed a crosswalk between Federal job classifications and the new Standard Occupational Classification System.  BLS used the new crosswalk and March 2001 GS employment weights for data delivered this year.  OPM staff made seven changes in the crosswalk developed by the interagency group to match GS jobs to more specific SOC jobs.  These changes involved using the non-Federal employment distribution as measured by the Occupational Employment Survey conducted by BLS to allocate GS employment where the Government does not have detailed information.  For example, the interagency group matched all Federal computer programmers to the SOC Computer Specialists, All Other, job because we do not have detailed information on specialty occupations.  Unfortunately, BLS matches most of the data it collects to the specialty occupations-- Computer Programmer, Software Engineers, Systems Analysts, Database Administrators, Network Administrators, and Data Communications Analysts.  Similar modifications of the crosswalk were made for human resources specialists, drafters, engineering technicians, dentists, doctors, and secretaries.  We anticipate that additional improvements could be made in the crosswalk and that OPM will provide updated GS employment information to BLS each year.

5) Excluding randomly selected non-Federal jobs that would be classified above GS-15 in the Government.

Progress:  BLS has developed methods for identifying and excluding non-Federal jobs that would be classified above GS-15.  These data were excluded from the data delivered to the Pay Agent this year.

While we have recommended the phase-in of NCS data beginning with locality payments in 2004, we continue to have concerns about the reliability of data produced by the NCS.  As a result, we recommend that any further implementation of the use of NCS data in the calculation of locality pay gaps be conditioned on BLS addressing the following issues:

  1. Continued implementation by BLS of previously identified improvements to the NCS methodology:  At this point, BLS has implemented only three of the five improvements previously identified by the Federal Salary Council as necessary for use of NCS for locality pay calculations.  Although the Council has recommended partial use of the NCS data based on the implementation of these three improvements, any further implementation of (or additional weight given to) the NCS data must be expressly conditioned on action by BLS to implement the remaining two improvements identified by the Council.
  1. Sample Size:  The NCS relies heavily on modeling to provide "missing" data, i.e. data that is not collected from actual surveys because of insufficient matches in certain job categories.  For example, as noted in our Working Group report, "NCS random samples may miss key high-paying jobs that are not common in non-Federal establishments."  While modeling is an appropriate technique in the absence of available data, heavy reliance on modeling is troubling, especially when the modeled NCS data indicate a lower pay gap in a locality than had been identified using the former OCSP survey methodology.  In order to address this concern, additional resources should be committed by BLS to increase the sample size of its surveys, particularly in those localities where the NCS data indicate a pay gap that is more than 5 percentage points below the gap measured using OCSP data, or, at a minimum, conducting "augmentation surveys" for areas under-represented in BLS samples.  This will not only greatly increase the reliability of the NCS data, but also its credibility with Federal employees, Congress, and the public in general.

Allocating Locality Pay in 2003

In the past, the Council has recommended and the President has agreed to allocate funds available for locality pay raises based on the size of the pay gap in each area.  We recommend that funds available for locality pay in 2003 be allocated as follows:

Instead of applying a uniform phase-in factor--across-the-board--to all localities, the Pay Agent should base increases on the size of the pay gap in each locality, so that areas with bigger gaps than the average target gap (27.57 percent based on 2001 pay gaps for 2003 payments) would get bigger increases than those resulting from application of the uniform phase-in factor, while areas with smaller gaps than the average would get smaller increases.

At this point, we do not know what amount will be provided for 2003.  Presently, the Congress has draft appropriations bills that would set the total increase at 4.1 percent.  If that amount is approved, the Council recommends that 3.1 percent be allocated for the across-the-board pay raise and that 1.0 percent be allocated for locality pay raises.  Attachment 2 shows the rates for each area under this recommendation. 

Future Plans

Our Working Group plans to meet again soon to begin planning for the 2003 review of the structure of the locality pay program.  In 2003, we will have the new census data and new metropolitan area definitions and plan to consider other aspects of the program at that time, including how many locality pay areas are feasible, what constitutes a meaningful difference in locality rates, the relative precision of the pay gaps and locality rates, and how to treat areas that cannot be surveyed separately.  We will automatically review at that time all the areas that have contacted us about becoming areas of application.  

By direction of the Council:

                                                                                    Samuel J. Wallace

                                                                                    Chairman

Attachments