This is the accessible text file for GAO report number GAO-12-564 
entitled 'Federal Workers: Results of Studies on Federal Pay Varied 
Due to Differing Methodologies' which was released on July 23, 2012. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as 
part of a longer term project to improve GAO products' accessibility. 
Every attempt has been made to maintain the structural and data 
integrity of the original printed product. Accessibility features, 
such as text descriptions of tables, consecutively numbered footnotes 
placed at the end of the file, and the text of agency comment letters, 
are provided but may not exactly duplicate the presentation or format 
of the printed version. The portable document format (PDF) file is an 
exact electronic replica of the printed version. We welcome your 
feedback. Please E-mail your comments regarding the contents or 
accessibility features of this document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

United States Government Accountability Office: 
GAO: 

Report to Congressional Requesters: 

June 2012: 

Federal Workers: 

Results of Studies on Federal Pay Varied Due to Differing 
Methodologies: 

GAO-12-564: 

GAO Highlights: 

Highlights of GAO-12-564, a report to congressional requesters. 

Why GAO Did This Study: 

A careful consideration of federal pay is an essential part of fiscal 
stewardship and is necessary to support the recruitment and retention 
of a competent, successful workforce. Recent studies comparing the 
compensation of federal workers to workers in other sectors have 
produced varying findings. To improve understanding of federal pay 
setting, GAO was asked to examine (1) how annual pay adjustments for 
the GS system are determined; (2) the extent to which the pay 
increases and awards available to GS employees recognize individual 
performance, and how the Office of Personnel Management (OPM) provides 
oversight of pay increases and awards; and (3) how selected studies 
compare federal and private pay and total compensation and the factors 
that may account for the different findings. 

GAO reviewed legislation, OPM regulations, executive orders, and 
federal agency documents; analyzed OPM data; and interviewed agency 
officials. GAO reviewed six studies that met three criteria: issuance 
since 2005, original analysis, and focus on federal and private sector 
compensation. GAO compared and contrasted the differences between 
their approaches, methodologies, and data sources, and interviewed the 
studies’ authors, people with expertise in compensation issues, and 
agency officials responsible for the data. GAO provided drafts to 
agencies and study authors for review and comment and made technical 
changes as appropriate in response to comments received. One study 
author provided written comments concurring with the findings. 

What GAO Found: 

Annual pay adjustments for the General Schedule (GS), the pay system 
covering the majority of federal workers, are either determined 
through the process specified in the Federal Employees Pay 
Comparability Act of 1990 (FEPCA) or set based on percent increases 
authorized directly by Congress. GS employees receive an across-the-
board increase (ranging from 0 to 3.8 percent since FEPCA was 
implemented) that has usually been made in accordance with a FEPCA 
formula linking increases to national private sector salary growth. 
This increase is the same for each employee. GS employees also receive 
a locality increase that varies based on their location; 
there were 34 pay localities in 2012. While FEPCA specifies a process 
designed to reduce federal-nonfederal pay gaps in each locality, in 
practice locality increases have usually been far less than the 
recommended amount, which has been over 15 percent in recent years. 
For 2012, when there was a freeze on annual pay adjustments, the FEPCA 
process had recommended a 1.1 percent across-the-board increase and an 
average 18.5 percent locality increase. 

GS employees are eligible to receive three types of pay increases and 
monetary awards that are linked to individual performance appraisals: 
within-grade increases, ratings-based cash awards, and quality step 
increases. Within-grade increases are the least strongly linked to 
performance, ratings-based cash awards are more strongly linked to 
performance depending on the rating system the agency uses, and 
quality step increases are also more strongly linked to performance. 

Findings of selected pay and total compensation (pay and benefit) 
comparison studies varied due to different approaches, methods, and 
data. Regarding their pay analysis, the studies’ conclusions varied on 
which sector had the higher pay and the size of pay disparities. 
However, the overall pay disparity number does not tell the whole 
story; each of the studies that examined whether differences in pay 
varied among categories of workers, such as highly or less educated 
workers or workers in different occupations, found such variations. 
Three approaches were used to compare pay: 
* human capital approach (3 studies)—compares pay for individuals with 
various personal attributes (e.g., education, experience) and other 
attributes (e.g., occupation, firm size); 
* job-to-job approach (2 studies)—compares pay for similar jobs of 
various types based on job-related attributes such as occupation, does 
not take into account the personal attributes of the workers currently 
filling them; and; 
* trend analysis approach (1 study)—illustrates broad trends in pay 
over time without controlling for attributes of the workers or jobs. 

When looking within and across the studies, it is important to 
understand the studies’ differences in approach, methods, and data 
because they impact how the studies can be interpreted. The 
differences among the selected studies are such that comparing their 
results to help inform pay decisions is potentially problematic. Given 
the different approaches of the selected studies, their findings 
should not be taken in isolation as the answer to how federal pay and 
total compensation compares with other sectors. 

What GAO Recommends: 

GAO is not making any recommendations in this report. 

View [hyperlink, http://www.gao.gov/products/GAO-12-564]. For more 
information, contact Robert Goldenkoff at (202) 512-6806 or 
goldenkoffr@gao.gov. 

[End of section] 

Contents: 

Letter: 

Background: 

GS Pay Adjustments Can Be Determined through FEPCA Process or Set 
Directly by Congress: 

Pay Increases and Awards for GS Employees Vary in their Links to 
Individual Performance: 

Findings of Selected Pay and Total Compensation Comparison Studies 
Varied Due to Different Approaches, Methods, and Data: 

Concluding Observations: 

Agency and Third-Party Comments: 

Appendix I: Objectives, Scope, and Methodology: 

Appendix II: Implementation of Locality Pay: 

Appendix III: Information on the Selected Studies' Data Sources and 
Methodologies for Analyzing Pay and Total Compensation: 

Appendix IV: Comments from the Project On Government Oversight: 

Appendix V: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Eligibility Requirements for Ratings-Based Pay Increases and 
Awards for GS Employees: 

Table 2: Selected Studies' Findings on Pay: 

Table 3: Approaches Used by Selected Studies in Analyzing Pay: 

Table 4: Comparisons among the Selected Studies' Methodologies: 

Table 5: Selected Studies' Findings on Total Compensation (Pay and 
Benefits): 

Table 6: Selected Studies' Data and Methodologies Used in Analyzing 
Pay: 

Table 7: Selected Studies' Data and Methodologies Used in Analyzing 
Total Compensation (Pay and Benefits): 

Table 8: Data Sources Used in Analyzing Compensation: 

Figures: 

Figure 1: Annual Pay for a GS-11, Step 1 Employee in Selected Pay 
Localities, 2012: 

Figure 2: Congress and the President Established Across-the-Board and 
Locality Pay Adjustments, 2007 to 2012: 

Figure 3: Average Ratings-Based Award Amounts in Fiscal Year 2011, by 
Rating, 5-level Rating System: 

Figure 4: Ratings-Based Pay Increases and Awards for GS Employees for 
Fiscal Year 2011: 

Figure 5: Annual Pay for a GS-11 Employee in San Francisco and Rest of 
U.S. Pay Localities and Reported Nonfederal Equivalents, 1994 to 2011: 

Figure 6: Surveys and Models for Calculating Locality Increases Have 
Changed Several Times from 1990 to 2011: 

Abbreviations: 

BEA: Bureau of Economic Analysis: 

BLS: Bureau of Labor Statistics: 

CBO: Congressional Budget Office: 

CPDF: Central Personnel Data File: 

CPS: Current Population Survey: 

ECI: Employment Cost Index: 

EHRI-SDM: Enterprise Human Resources Integration-Statistical Data Mart: 

FEPCA: Federal Employees Pay Comparability Act of 1990: 

GL: GS pay plan that covers law enforcement officers who receive 
special base rates at grades 3-10 under section 403 of FEPCA: 

GM: GS pay plan that covers employees covered by the Performance 
Management and Recognition System termination provisions: 

GP: GS pay plan that covers GS physicians and dentists paid market pay: 

GR: GS pay plan that covers physicians and dentists covered by the 
Performance Management and Recognition System termination provisions 
paid market pay: 

GS: General Schedule: 

NCS: National Compensation Survey: 

NIPA: national income and product accounts: 

OCSP: Occupational Compensation Survey Program: 

OES: Occupational Employment Statistics: 

OMB: Office of Management and Budget: 

OPM: Office of Personnel Management: 

PATCO: Professional, Administrative, Technical, Clerical, and Other 
White-Collar: 

POGO: Project On Government Oversight: 

PPA: President's Pay Agent: 

[End of section] 

United States Government Accountability Office: 
Washington, DC 20548: 

June 22, 2012: 

The Honorable Darrell Issa: 
Chairman: 
Committee on Oversight and Government Reform: 
House of Representatives: 

The Honorable Dennis Ross: 
Chairman: 
Subcommittee on Federal Workforce, U.S. Postal Service and Labor 
Policy: 
Committee on Oversight and Government Reform: 
House of Representatives: 

Skilled federal workers are integral to the successful operation of 
every government function. They are law enforcement officers, 
engineers, program managers, scientists, clerks, and all the other 
workers who carry out the federal government's business. In 2011, 
there were approximately 2.2 million federal civilian workers 
comprising about 1.6 percent of the U.S. workforce, and civilian 
personnel costs were about $220 billion, comprising about 6 percent of 
the total federal budget.[Footnote 1] A careful consideration of 
federal workers' pay is an essential part of fiscal and operational 
stewardship and is necessary to support the recruitment and retention 
of a competent, successful workforce. It is the policy of Congress 
that pay for federal workers in the General Schedule (GS), the pay 
system covering the majority of federal workers,[Footnote 2] be in 
line with pay for comparable nonfederal workers. 

The composition of the federal workforce has changed over the past 30 
years, with the need for clerical and blue collar roles diminishing 
and professional, administrative, and technical roles increasing. As a 
result, today's federal jobs require more advanced skills at higher 
grade levels than federal jobs 30 years ago. Additionally, federal 
jobs, on average, require more advanced skills and degrees than 
private sector jobs. This is because a higher proportion of federal 
jobs than nonfederal are in skilled occupations such as science, 
engineering, and program management, while a lower proportion of 
federal jobs than nonfederal are in occupations such as manufacturing, 
construction, and service work. The result is that the federal 
workforce is on average more highly educated than the private sector 
workforce. 

Given the changes in the federal workforce over the last 30 years, 
there has been growing interest in reexamining the federal pay system--
how pay is determined and how comparisons with other sectors are made. 
Specifically, we have reported on the importance of considering the 
skills, knowledge, and performance of employees as well as the local 
labor market in making pay decisions.[Footnote 3] The President's Pay 
Agent, the entity responsible for recommending federal locality pay 
adjustments to the President, has recommended that the underlying 
model and methodology for estimating pay gaps be reexamined to ensure 
that private sector and federal sector pay comparisons are as accurate 
as possible. 

The Subcommittee on Federal Workforce, U.S. Postal Service and Labor 
Policy held a hearing in March 2011 on the issue of comparability in 
pay and benefits between the federal and nonfederal workforces. 
[Footnote 4] The research presented at the hearing showed varying 
findings on the gap in pay between federal and private sector workers 
as estimated by outside organizations as well as the gap between 
federal and nonfederal workers as estimated by the President's Pay 
Agent. The outside organizations' studies also varied in their 
findings on the gap in total compensation (pay plus benefits). 
[Footnote 5] The hearing raised questions about how the methodologies 
used by the various authors differed and how these differences 
affected their overall findings. 

Accordingly, at your request, this report examines (1) how annual pay 
adjustments for the GS system are determined; (2) the extent to which 
the pay increases and awards available to GS employees recognize 
individual performance, and how the Office of Personnel Management 
(OPM) provides oversight of pay increases and awards; and (3) how 
selected studies compare federal and private sector pay and total 
compensation and the factors that may account for the different 
findings. 

To examine how GS annual across-the-board and locality pay adjustments 
are determined, we reviewed legislation, OPM regulations, executive 
orders, Presidents' alternative pay plans, President's Pay Agent 
Reports, Federal Salary Council recommendations, OPM and Bureau of 
Labor Statistics (BLS) documents and reports, and reports by the 
Congressional Budget Office (CBO) and Congressional Research Service. 
[Footnote 6] We also examined how the methodology for determining 
locality pay has changed since the start of locality pay to the 
present. We interviewed selected members of the Federal Salary Council 
and its working group; the Council is to be made up of six 
representatives of federal employee groups and three experts in labor 
relations, and makes annual recommendations to the President's Pay 
Agent.[Footnote 7] We interviewed BLS officials, OPM officials who are 
knowledgeable about federal pay policy and serve as staff to the 
President's Pay Agent, and people with expertise in compensation 
issues including former federal officials experienced with pay and 
benefits issues. 

To determine the extent to which pay increases and awards recognize 
individual performance, we analyzed legislation and OPM regulations on 
pay increases and awards available to employees in the GS pay system 
and identified those pay increases and awards that are determined in 
part by an individual's performance rating as measured by the agency's 
performance appraisal system. We analyzed data for employees in the GS 
pay plan in the aggregate on the number, percentage, and dollar amount 
of selected pay increases and awards; the amount of these increases 
and awards as a portion of GS payroll (total adjusted basic pay for 
all employees in the GS pay plan); and the distribution of these 
increases and awards by rating pattern and rating levels from OPM's 
Central Personnel Data File (CPDF) for fiscal year 2011. To help 
determine the reliability and accuracy of the CPDF data elements used, 
we checked the data for reasonableness and the presence of any obvious 
or potential errors in accuracy and completeness; reviewed past GAO 
analyses of the reliability of CPDF data; and interviewed OPM 
officials knowledgeable about the data. We believe the CPDF is 
sufficiently reliable for the purpose of this report. To describe how 
OPM provides oversight of pay increases and awards, we collected and 
analyzed OPM guidance to agencies including regulations, memoranda, 
reports, fact sheets, and frequently asked questions. We also 
interviewed OPM officials responsible for federal pay policies to 
discuss the implementation of the guidance and monitoring of agencies' 
use of increases and awards through reports and other means; 
and interviewed OPM officials responsible for conducting human capital 
management evaluations at agencies on pay increases and awards and the 
overall GS pay system, among other things. 

To review selected studies that compare federal and private sector pay 
and total compensation and describe factors that help account for the 
different study findings, we conducted a detailed literature review of 
academic journals, agency and organization publications, and other 
sources and applied three criteria to the results, selecting five 
studies that: (1) were published/issued since 2005; (2) include 
original analysis; and (3) have the explicit and primary purpose of 
comparing federal and private sector pay and total compensation. 
[Footnote 8] After our literature review was completed, CBO issued a 
study comparing federal and private sector compensation. We included 
this study in our review because it met our criteria. That brought the 
total number of selected studies up to six. We did not examine the 
reliability or the appropriateness of the approaches, methods, and 
data used by the studies, and we did not exclude any study on the 
basis of methodological quality. We reviewed the studies, summarized 
each study's methodologies and key findings, and confirmed the 
accuracy of our summaries with the authors. We compared and contrasted 
the differences between the approaches, methodologies, and data 
sources of the selected studies. We interviewed the selected study 
authors to obtain their views on the various methodologies and data 
sources available, why they chose the ones they used, and their 
conclusions based on their work. We interviewed a number of 
individuals chosen for their expertise in compensation issues to 
obtain their views on the data sources for analyzing compensation and 
to provide a general context for the issues involved in comparing 
federal and private or nonfederal pay and total compensation. The 
findings regarding the selected studies are not based on input from 
these individuals. We interviewed officials from the Bureau of 
Economic Analysis (BEA), BLS, and the U.S. Census Bureau to discuss 
how these agencies' data are used to measure federal and private or 
nonfederal pay, compensation, or benefits, and limitations of their 
data or surveys. We also interviewed officials from OPM involved in 
federal pay policies. (See appendix I for a more detailed discussion 
of our objectives, scope, and methodology.) 

We conducted this performance audit from July 2011 to June 2012 in 
accordance with generally accepted government auditing standards. 
Those standards require that we plan and perform the audit to obtain 
sufficient, appropriate evidence to provide a reasonable basis for our 
findings and conclusions based on our audit objectives. We believe 
that the evidence obtained provides a reasonable basis for our 
findings and conclusions based on our audit objectives. 

Background: 

The GS pay system covered 69 percent of federal civilian workers in 
2011, with compensation costing about $147 billion (about 67 percent 
of total federal civilian compensation of about $220 billion). 
[Footnote 9] The GS workforce is divided into 15 pay grades, with 10 
rates of pay (referred to as steps) within each grade. Agencies use a 
uniform set of classification standards to determine grade levels for 
their positions organized within five occupational categories--
Professional, Administrative, Technical, Clerical, and Other White-
Collar (PATCO). The GS system of classification was established by the 
Classification Act of 1949 in response to calls for a modernized 
system to ensure equity in pay setting.[Footnote 10] Until the late 
1960s, general pay adjustments for federal employees were made through 
acts of Congress.[Footnote 11] The Federal Pay Comparability Act of 
1970 permanently authorized the President to adjust GS[Footnote 12] 
pay rates annually, and established a system for recommending 
adjustments with the goal of increasing federal pay to be comparable 
with the private sector; however, we previously found that the gap 
between average federal and private sector salaries for similar jobs 
continued after implementation of the act because the recommended 
adjustments were not always made.[Footnote 13] 

The Federal Employees Pay Comparability Act of 1990 (FEPCA) created 
annual locality-based pay adjustments for GS employees to reduce 
reported gaps between federal and nonfederal pay in metropolitan 
areas.[Footnote 14] In addition, FEPCA maintained an annual across-the-
board pay adjustment that is the same for each employee to keep the GS 
base pay schedule in line with salary growth in the general labor 
market, similar to what had already existed under the 1970 act. 
[Footnote 15] 

Before FEPCA, federal employees doing the same job at the same level 
anywhere in the country were paid the same amount.[Footnote 16] 
However, there was a growing concern that it was difficult to recruit 
and retain skilled federal employees in areas with higher nonfederal 
wages. We concluded that locality-based pay adjustments were 
necessary.[Footnote 17] FEPCA established locality pay, and the 
President's Pay Agent designated pay localities based on Office of 
Management and Budget (OMB) Metropolitan Statistical Areas.[Footnote 
18] FEPCA's goal was to reduce the gap between federal and nonfederal 
pay in each locality, as measured by BLS data and reported by the 
President's Pay Agent, to 5 percent over the course of 9 years. This 
goal was not met, but locality pay increases have been provided every 
year since locality pay was implemented in 1994, except during the pay 
freeze in 2011 and 2012.[Footnote 19] According to OPM, locality pay 
is now a broadly accepted practice in federal pay administration. See 
appendix II for more information on the implementation of locality pay. 

Figure 1 illustrates the extent to which locality pay has been 
implemented for a representative employee. The annual pay (base plus 
locality) in 2012 for an employee at GS-11 (approximately the midpoint 
grade level), step 1 is shown for selected pay localities. Examples of 
positions that a GS-11 employee might hold are Administrative Officer, 
Scientist, Paralegal Specialist, Accountant, Engineer, Medical Records 
Administrator, Nurse Specialist, and Information Technology 
Specialist. There were 34 pay localities in the United States in 2012, 
composed of the states of Alaska and Hawaii, 31 metropolitan areas, 
and a residual locality called "Rest of U.S." that includes all other 
areas in the United States and its territories and possessions. Rest 
of U.S. was the lowest-paying locality in 2012, with a GS-11, step 1 
earning $57,408, and San Francisco was the highest, with a GS-11, step 
1 earning $67,963. We selected additional localities with various pay 
rates and population sizes and from various regions to create figure 1. 

Figure 1: Annual Pay for a GS-11, Step 1 Employee in Selected Pay 
Localities, 2012: 

[Refer to PDF for image: stacked vertical bar graph] 

Locality pay area: Rest of U.S. 
Base pay: $50,287; 
Locality pay: $7,121; 
Total: $57,408. 

Locality pay area: Huntsville; 
Base pay: $50,287; 
Locality pay: $8,056; 
Total: $58,343. 

Locality pay area: Columbus; 
Base pay: $50,287; 
Locality pay: $8,629; 
Total: $58,916. 

Locality pay area: Dallas; 
Base pay: $50,287; 
Locality pay: $10,394; 
Total: $60,682. 

Locality pay area: Minneapolis; 
Base pay: $50,287; 
Locality pay: $10,540; 
Total: $60,827. 

Locality pay area: Washington, D.C. 
Base pay: $50,287; 
Locality pay: $12,180; 
Total: $62,467. 

Locality pay area: New York; 
Base pay: $50,287; 
Locality pay: $14,442; 
Total: $64,729. 

Locality pay area: San Francisco; 
Base pay: $50,287; 
Locality pay: $17,676; 
Total: $67,963. 

Source: GAO analysis of OPM data. 

[End of figure] 

GS Pay Adjustments Can Be Determined through FEPCA Process or Set 
Directly by Congress: 

Across-the-Board Adjustments Are Usually Based on Private Sector 
Salary Growth: 

Across-the-board adjustments are designed to keep the GS base pay 
schedule in line with salary growth in the general labor market. FEPCA 
specifies that unless the President provides for alternative pay 
adjustments, across-the-board pay adjustments are to be determined 
using a simple formula: Pay rates are to be increased by the 12-month 
percentage increase in the wage and salary component of the Employment 
Cost Index (ECI) for private sector workers, minus one-half of one 
percentage point. For example, the ECI reference period for the 
January 2013 increase is the 12-month period ending September 2011. 
The ECI shows that during that period, pay for private sector workers 
rose by 1.7 percent.[Footnote 20] Therefore, the across-the-board 
increase for 2013 would be 1.2 percent. The ECI, an index compiled by 
the BLS and published quarterly, measures percentage changes in wages 
and salaries for private sector employees. 

As specified in FEPCA, the President may decide to either provide 
across-the-board pay adjustments based on this calculation, or provide 
alternative pay adjustments based on national emergency or serious 
economic conditions affecting the general welfare.[Footnote 21] 
Additionally, Congress may legislate an increase that is different 
from the formula result or the President's alternative plan; 
this is not part of the process specified by FEPCA. The FEPCA formula 
increase has gone into effect in 12 of the 19 years since 1994; 
the largest increase was 3.8 percent.[Footnote 22] An amount lower 
than the formula amount went into effect in the other 7 years due to 
President's alternative pay plans and laws passed by Congress. 
[Footnote 23] The smallest increase was 0 percent, during the freeze 
on annual pay adjustments in 2011 and 2012.[Footnote 24] 

Locality Adjustments Are Usually Set by Congress: 

Locality adjustments are designed to reduce the gap between federal 
and nonfederal pay in each locality to no more than 5 percent based on 
surveys to be conducted by BLS. FEPCA specifies that locality pay 
adjustments are to be recommended by a Pay Agent designated by the 
President, which is to consider the views of employee organizations: 

* The President's Pay Agent recommends annual comparability payment 
amounts, establishes and modifies pay localities as it considers 
appropriate, and submits an annual report to the President on these 
items. The Secretary of Labor and the Directors of OMB and OPM serve 
as the Pay Agent. In making its recommendations, the President's Pay 
Agent considers the views and recommendations of a Federal Salary 
Council and other employee organizations. 

* The Federal Salary Council makes annual recommendations to the 
President's Pay Agent on locality pay adjustments, including the 
establishment or modification of pay localities, the coverage of 
salary surveys used to set locality pay, the process for making pay 
comparisons, and the level of comparability payments that should be 
made. The Council is to be comprised of three experts in labor 
relations and pay policy and six representatives of employee 
organizations representing large numbers of GS employees.[Footnote 25] 

[Sidebar: Current Process for Comparing Federal and Nonfederal Pay: 

The Pay Agent is charged with determining the disparities between 
federal and nonfederal pay in each locality. It measures federal pay 
based on OPM records that identify GS employees by occupation and 
grade level, and nonfederal pay based on BLS data The Pay Agent asks 
BLS to provide a statistical estimate of nonfederal pay rates for each 
locality by GS grade level and occupation (with occupations classed in 
five broad occupational groups - Professional, Administrative, 
Technical, Clerical, and Other White-Collar). BLS provides these 
estimates based on its National Compensation Survey (NCS), using 
statistical modeling to fill in some of the data. 

The Pay Agent averages these pay rates together, weighting the 
averages by equivalent federal GS employment levels. For example, if
50 percent of federal workers in a given locality have Administrative 
jobs, then the average pay for nonfederal Administrative workers in 
that locality are weighted to account for 50 percent of the pay 
estimate for that locality. This is to ensure that the Pay Agent is 
comparing "apples to apples," i.e. comparing similar types of jobs 
with each other. The Pay Agent compares the resulting average 
nonfederal pay estimate for each locality with the average federal pay 
for each locality to arrive at the pay disparities. The
Pay Agent bases its locality pay increase recommendations to the 
President on these disparities. End of sidebar] 

To recommend locality pay adjustments, the President's Pay Agent 
compares the annual GS base pay rates of federal workers in each area 
to the annual pay rates of nonfederal workers in the same area for the 
same levels and types of work. The sidebar provides details on this 
process. The target locality pay is the amount that reduces these 
differences to 5 percent. 

The surveys and models used for making these pay comparisons have 
changed somewhat between the passage of FEPCA in 1990 and the 2011 
President's Pay Agent Report (which recommends pay adjustments for 
2013 and is the most current report available). Some changes were 
initiated by BLS, and some changes were made in response to concerns 
expressed by the Federal Salary Council or President's Pay Agent. For 
example, BLS changed the survey used to measure nonfederal pay in 1996; 
the Federal Salary Council and President's Pay Agent expressed 
concerns, and BLS worked together with OPM and OMB to improve the 
suitability of the new survey for recommending locality payments. 
Improvements were phased in from 2002 to 2011. Changes are summarized 
in appendix II. 

As specified in FEPCA and similar to the process for across-the-board 
adjustments, the President may decide to either provide locality pay 
adjustments based on the Pay Agent's recommendation, or provide for 
alternative pay adjustments based on national emergency or serious 
economic conditions affecting the general welfare. Additionally, 
Congress may legislate an average percent increase that is different 
from the Pay Agent's recommendation or the President's alternative 
plan; this is not part of the process specified by FEPCA. 

For 1994, the first year that locality payments were made, FEPCA 
specified that the locality increase should be not less than one fifth 
of the amount needed to reduce the pay disparity to 5 percent. 
[Footnote 26] This amount, providing a 3.95 percent average locality 
pay rate for the average GS employee as recommended by the Pay Agent, 
went into effect. In subsequent years through 2012, the effective 
increase has usually been far less than the one recommended by the Pay 
Agent, either due to a President's alternative pay adjustment or to a 
law passed by Congress.[Footnote 27] 

Nonetheless, some locality pay increase has been provided every year 
since locality pay was implemented in 1994 (except during the pay 
freeze in 2011 and 2012), and reported disparities between federal and 
nonfederal pay by locality have been reduced. The President's Pay 
Agent reported that pay disparities were lower in 2011 than in 1994 in 
16 of the 21 pay localities that existed in both of those years. 
Federal Salary Council members and OPM officials we spoke with said 
that FEPCA was successful in its goal of improving federal pay setting 
for large metropolitan areas by more closely aligning pay to local 
labor markets. 

Figure 2 summarizes pay adjustments during the past 6 years, 
illustrating the differences between the President's Pay Agent 
recommendations and the final effective amounts. These differences 
were driven primarily by locality pay, since the across-the-board 
adjustments required under the FEPCA formula were smaller and were 
provided in some years, while the recommended locality adjustments 
were larger and were not provided. For example, for 2007, the 
President's Pay Agent recommended a 1.7 percent across-the-board 
increase to comply with the formula in FEPCA, and a 7 percent average 
locality increase based on BLS salary survey data. The President 
provided for the 1.7 percent across-the-board increase but limited the 
average locality increase to the alternative amount of 0.5 percent. 

As another example, for 2012, the FEPCA process specified a 1.1 
percent across-the-board increase and an average 18.5 percent locality 
increase, but annual pay adjustments were frozen instead.[Footnote 28] 
The Pay Agent had reported that in 2010 (the reference year for 
setting 2012 pay), taking both across-the-board and locality pay into 
account, the average federal-nonfederal pay gap was 24 percent. The 
approximately 20 percent overall average increase recommended by the 
Pay Agent for 2012 would have lowered the pay disparity to FEPCA's 
target of 5 percent. 

Figure 2: Congress and the President Established Across-the-Board and 
Locality Pay Adjustments, 2007 to 2012: 

[Refer to PDF for image: horizontal bar graph] 

Year: 2007; 

President's Pay Agent recommendations: 
Across-the-board: 1.7%; 
Average locality increase[A]: 7.0%. 
President's alternative: 
Across-the-board: 1.7%; 
Average locality increase[A]: 0.5%. 
Congress's decision (statute)[B]: [Empty]. 
Final effective amount: 
Across-the-board: 1.7%; 
Average locality increase[A]: 0.5%. 

Year: 2008; 

President's Pay Agent recommendations: 
Across-the-board: 2.5%; 
Average locality increase[A]: 12.4%. 
President's alternative: 
Across-the-board: 2.5%; 
Average locality increase[A]: 0.5%. 
Congress's decision (statute)[B]: 3.5%[B]. 
Final effective amount: 
Across-the-board: 2.5%; 
Average locality increase[A]: 1%. 

Year: 2009; 
President's Pay Agent recommendations: 
Across-the-board: 2.9%; 
Average locality increase[A]: 16.3%. 
President's alternative: [Empty]. 
Congress's decision (statute)[B]: 3.9%[B]. 
Final effective amount: 
Across-the-board: 2.9%; 
Average locality increase[A]: 1%. 

Year: 2010; 
President's Pay Agent recommendations: 
Across-the-board: 2.4%; 
Average locality increase[A]: 16.6%. 
President's alternative: 
Across-the-board: 2.0%; 
Average locality increase[A]: 0%. 
Congress's decision (statute)[B]: 
Across-the-board: 1.5%; 
Average locality increase[A]: 0.5%. 
Final effective amount: 
Across-the-board: 1.5%; 
Average locality increase[A]: 0.5%. 

Year: 2011; 
President's Pay Agent recommendations: 
Across-the-board: 0.9%; 
Average locality increase[A]: 15.9%. 
President's alternative: Plans recommends 2-year pay freeze. 
Congress's decision (statute)[B]: Statue enacts 2-year pay freeze. 
Final effective amount: Pay freeze. 

Year: 2012; 
President's Pay Agent recommendations: 
Across-the-board: 1.1%; 
Average locality increase[A]: 18.5%. 
President's alternative: [Empty]. 
Congress's decision (statute)[B]: 2nd year of pay freeze. 
Final effective amount: Pay freeze. 

Source: GAO analysis of legislation and President's Pay Agent, 
Executive Office of the President and OPM data. 

[A] Average locality increase: The average percentage by which a GS 
employee's salary would increase from the previous year due to 
locality pay. Employees in localities with below-average pay gaps 
would receive lower locality adjustments, and those in localities with 
above-average pay gaps would receive higher adjustments. 

[B] In some years, Congress specified an overall percentage increase, 
allowing the President to decide how much would be for locality and 
how much for across-the-board. 

[End of figure] 

Pay Increases and Awards for GS Employees Vary in their Links to 
Individual Performance: 

The pay increases and awards available to GS employees are designed to 
recognize individual performance to varying degrees. Across-the-board 
and locality pay increases, which are given to all covered employees 
nearly every year, are not linked to performance at all. Awards such 
as suggestion/invention awards and superior accomplishment awards are 
designed to recognize performance without being linked specifically to 
performance ratings.[Footnote 29] 

Three pay increases and monetary awards available to GS employees are 
linked to performance ratings as determined by agencies' performance 
appraisal systems: 

* Within-grade increases are periodic increases in a permanent 
employee's rate of basic pay from one step of a grade to the next 
higher step within the grade.[Footnote 30] 

* Ratings-based cash awards are lump sum cash payments that are 
designed to recognize performance.[Footnote 31] 

* Quality step increases are faster-than-normal step increases that 
are designed to recognize excellence in performance.[Footnote 32] 

Table 1 describes these pay increases' and awards' eligibility 
requirements, which include a certain level of performance, frequency 
limits, and agency-specified criteria.[Footnote 33] As outlined in OPM 
regulations, agencies' performance appraisal systems can have varying 
ranges of summary performance rating levels--ranging from a pass/fail 
system with two summary rating levels to a system with five summary 
rating levels.[Footnote 34] All the systems used by the agencies 
include level 3, "fully successful," which is the pass level for a 
pass/fail system. 

Table 1: Eligibility Requirements for Ratings-Based Pay Increases and 
Awards for GS Employees: 

Increase or award[A]: Within-grade increase; 
Performance rating required: At least "fully successful" (level 3); 
Frequency limit or waiting period: One to 3 years' waiting period 
(with longer waiting periods at higher step rates). A quality step 
increase does not reset the waiting period but can move an employee to 
a step rate with a longer waiting period; 
Agency-specified criteria: Not applicable. Agencies are required to 
provide within-grade increases to all eligible employees. 

Increase or award[A]: Ratings-based cash award; 
Performance rating required: "Fully successful" (level 3) or higher 
(larger awards for higher ratings); 
Frequency limit or waiting period: Frequency limit, 1 per year[B]; 
Agency-specified criteria: Yes, performance-related and other 
criteria, such as consideration of other recognition recently received. 

Increase or award[A]: Quality step increase; 
Performance rating required: The highest possible rating in the 
agency's system[C]; 
Frequency limit or waiting period: Waiting period, 1 year since the 
last quality step increase; 
Agency-specified criteria: Yes, performance-related and other 
criteria[C]. 

Source: GAO analysis of legislation, regulations, and OPM guidance. 

[A] GS employees may be eligible to receive within-grade and quality 
step increases as well as ratings-based cash awards in the same year 
depending on their agencies' criteria and the applicable waiting 
periods for within-grade increases. 

[B] Per OPM guidance, ratings-based cash awards recognize an 
employee's performance over an entire rating period. Generally, 
performance rating periods should be 12 months. 5 C.F.R. § 
430.206(a)(2). 

[C] Employees not covered by 5-level systems (where "outstanding" is 
the highest level) must demonstrate sustained performance of high 
quality significantly above the "fully successful" level as determined 
under performance-related criteria established by the agency. 5 CFR 
531.504(b). 

[End of table] 

In practice, based on our analysis of CPDF data from fiscal year 2011, 
the degree to which individual performance drove receipt of these pay 
increases and awards for employees in the GS pay plan varied. Of the 
three pay increases and awards we analyzed, within-grade increases 
were the least strongly linked to performance. Ratings-based cash 
awards were more strongly linked to performance depending on the 
rating system the agency used, and quality step increases were also 
more strongly linked to performance. 

Within-grade increases were the least strongly linked to performance 
of the three pay increases and awards we analyzed, in accordance with 
their design. As noted in table 1, agencies are required to provide 
within-grade increases to employees whose performance is at least 
"fully successful" and who have finished their waiting period. Over 99 
percent of employees in the GS pay plan[Footnote 35] received 
performance ratings at or above "fully successful" in fiscal year 
2011. Thirty-nine percent received within-grade increases, comprising 
nearly all the employees who completed their waiting period. 

Ratings-based cash awards were more strongly linked than within-grade 
increases to performance. All GS pay plan employees may receive 
ratings-based cash awards every year (unlike within-grade increases), 
so frequency limits are not a primary determinant of who receives 
them. In fiscal year 2011, the degree of linkage of awards with 
performance ratings varied by the type of appraisal system used by the 
agency. 

In fiscal year 2011, 81 percent of employees in the GS pay plan were 
covered by 5-level rating systems and other systems that allowed for 
distinctions between "fully successful" and higher levels of 
performance.[Footnote 36] Ratings-based cash awards for these 
employees were given at higher rates to employees with better 
performance. For example, for the 5-level system, which covered 63 
percent of GS employees, awards were given to 65 percent of employees 
with "outstanding" ratings, 58 percent of employees with ratings 
"between outstanding and fully successful," and 24 percent of 
employees with "fully successful" ratings. Along with the performance 
rating received, agency criteria were used to determine who received 
awards. As noted in table 1, an agency should identify any other 
criteria to be considered when making award recommendations and 
decisions, including any other awards or personnel actions that should 
be taken into consideration such as time off, a quality step increase, 
or a recent promotion. 

In accordance with OPM regulations, employees with higher ratings are 
to receive larger ratings-based awards, and award patterns reflected 
this distinction in 2011, as shown in figure 3. Employees who received 
"outstanding" ratings within the 5-level system received the largest 
awards. 

Figure 3: Average Ratings-Based Award Amounts in Fiscal Year 2011, by 
Rating, 5-level Rating System: 

[Refer to PDF for image: vertical bar graph] 

Rating: Fully successful; 
Amount: $686. 

Rating: Level between outstanding and fully successful; 
Amount: $1,272. 

Rating: Outstanding; 
Amount: $1,780. 

Source: GAO analysis of OPM data. 

[End of figure] 

In fiscal year 2011, about 19 percent of employees in the GS pay plan 
were covered by a pass/fail rating system or another system that did 
not allow for distinctions in performance above the "fully successful" 
level. Over 99 percent of employees in these systems received a "fully 
successful" rating in fiscal year 2011, while only 31 percent received 
a ratings-based award, meaning that most decisions not to provide 
awards were made based on other criteria than ratings. Performance 
ratings and agency criteria, including performance-related criteria, 
were used to determine who received awards. 

Quality step increases were also more strongly linked to performance 
than within-grade increases. As shown in table 1, GS employees must 
perform at their agency's highest possible level to be eligible to 
receive a quality step increase. About 49 percent of employees 
received the highest possible rating their agency's system allowed in 
fiscal year 2011.[Footnote 37] Of those employees, about 7 percent 
received a quality step increase. Unlike within-grade increases, the 
waiting period for quality step increases is 1 year for all employees, 
eliminating the waiting period as a primary determinant for receiving 
quality step increases; 
rather, decisions were made based on performance rating and agency 
criteria, including performance-related criteria. 

Figure 4 illustrates the percentage of employees receiving each type 
of increase and award, the average amounts of the increases and awards 
in dollars and as a percent of the recipient's pay, and the cost to 
the government of ratings-based pay increases and awards for GS 
employees for fiscal year 2011. 

Figure 4: Ratings-Based Pay Increases and Awards for GS Employees for 
Fiscal Year 2011: 

[Refer to PDF for image: illustrated table] 

Increase or award: Within-grade increase[A]; 
Percentage receiving increase or award: 39.1%; 
Average (mean) amount of increase or award as percentage of 
recipient's pay[B]: 3.0%; 
Average (mean) amount of increase or award in dollars[B]: $2,062; 
Total federal dollars spent for increase or award (millions, and as 
percentage of $108 billion GS system total payroll)[C]: $1,050.5; 1% 
of GS payroll. 

Increase or award: Ratings-based cash award[D]; 
Percentage receiving increase or award: 46.8%; 
Average (mean) amount of increase or award as percentage of 
recipient's pay[B]: 1.8%; 
Average (mean) amount of increase or award in dollars[B]: $1,432; 
Total federal dollars spent for increase or award (millions, and as 
percentage of $108 billion GS system total payroll)[C]: $872.9; 0.8% 
of GS payroll. 

Increase or award: Quality step increase; 
Percentage receiving increase or award: 3.4%; 
Average (mean) amount of increase or award as percentage of 
recipient's pay[B]: 2.8%; 
Average (mean) amount of increase or award in dollars[B]: $2,384; 
Total federal dollars spent for increase or award (millions, and as 
percentage of $108 billion GS system total payroll)[C]: $104.1; 
0.07% of GS payroll. 
             
Source: GAO analysis of OPM data.              

[A] Not all GS employees are eligible to receive a within-grade 
increase every year due to waiting periods dependent on the employee's 
step, as discussed in table 1. 

[B] Average amounts of increases or awards as percentages of base pay 
and in dollars are for employees receiving awards or increases only; 
zero values are not included in the averages. Average percentage 
increases, average dollars spent, and total dollars spent on within-
grade increases and quality step increases are annualized figures, 
i.e. full-year amounts are counted for increases that were given part 
of the way into the year. For within-grade increases and quality step 
increases, the percentage increase declines as the steps increase 
(except for grades 1 and 2). Average amount of increase or award in 
dollars and total dollars spent for increase or award are based on 
those records with complete award value data. 

[C] GS system total payroll is defined as aggregate adjusted basic pay 
for employees in the GS pay plan. This does not include benefits. 

[D] Ratings-based cash awards up to 10 percent of an employee's annual 
rate of pay are permitted. Agency heads may authorize an award 
exceeding 10 percent (but not in excess of 20 percent) if exceptional 
performance justifies such an award. 5 U.S.C. § 4505a. Awards in 
excess of $10,000 are to be submitted to OPM for approval. 5 U.S.C. § 
4502(b). 

[End of figure] 

OPM Provides Oversight on Awards and Increases: 

OPM's role with respect to awards and increases includes: 

* providing policy direction to agencies, including regulations, 

* reporting on agencies' use of awards and increases, and: 

* evaluating agencies' linkage of awards and increases with results. 

Agencies, in turn, must ensure they have met statutory and regulatory 
requirements and may develop agency-specific criteria for providing 
quality step increases and cash awards. According to OPM officials, 
awards regulations are highly decentralized because the statutes 
provide agency heads, not OPM, with the authority to grant awards. 
[Footnote 38] Likewise, agency heads may grant quality step increases 
within the limits of available appropriations and regulatory 
requirements. 

Policy direction. To help agencies understand how to administer pay 
increases and awards, OPM issues regulations and supporting memoranda 
and posts fact sheets, frequently asked questions, and other resource 
documents on its website. Topics have included approaches to 
calculating ratings-based cash awards, tax issues for awards, how the 
timing of quality step increases affects within-grade increases, and 
recent limitations on awards given budgetary constraints. According to 
OPM officials, OPM responds to agency questions about guidance as 
needed. 

Reporting. OPM provides agencies with an annual Federal Award 
Statistics report on cash awards, time-off awards, quality step 
increases, and other awards received by GS and other employees. 
[Footnote 39] According to OPM officials, OPM uses the report to show 
trends and compare usage of awards between agencies and across the 
government. OPM also uses the report data to help inform its decisions 
about awards policy and monitor agency compliance with the policy, 
such as limitations on awards usage. 

Evaluation. OPM evaluates selected agencies' human capital management 
systems as part of its broader strategy for maintaining human capital 
accountability. As part of these evaluations, OPM determines whether 
an agency's human capital system provides and clearly communicates 
linkages between employee performance expectations, performance 
recognition through increases and awards, and the agency's mission. 
OPM also reviews a sample of case files to check that the awards 
granted meet the requirements of the law and regulations,[Footnote 40] 
and assists agencies in leading their own evaluations. 

OPM officials said that they have identified the following issues in 
regard to pay increases and awards: 

* Some agencies tried to circumvent limitations on award amounts by 
issuing several incremental awards within a short time period. 

* Some agencies granted quality step increases to compensate for low 
award budgets. 

* Some agencies' human capital management systems did not link 
individual performance expectations and recognition through pay 
increases and awards to the accomplishment of specific mission-related 
goals or milestones. 

When OPM determines that an agency violated the law or regulations, 
such as circumventing award limitations by issuing several incremental 
awards within short periods of time, it requires the agency to take 
corrective action and respond to OPM with evidence of how it addressed 
or plans to address the violation within 60 days. For example, 
according to an OPM official, corrective action may result in the 
agency recovering the award from the recipient and correcting the 
documentation for the award. 

When OPM observes an issue with an agency's award implementation that 
does not violate regulations, OPM may recommend to the agency 
improvements that could be made. For example, when OPM determines that 
an agency has granted quality step increases to compensate for a low 
award budget, it recommends that the agency review its policies for 
granting pay increases and awards to ensure the policies comply with 
the intent of the laws and regulations. According to an OPM official, 
OPM requires an agency to respond to the recommendations made, but the 
agency is not required to take action on addressing the issue. 

Findings of Selected Pay and Total Compensation Comparison Studies 
Varied Due to Different Approaches, Methods, and Data: 

Selected Studies Differed in Their Conclusions and Basic Approaches to 
Analyzing Pay: 

The different study designs used by the authors of six studies 
resulted in varying conclusions on how federal pay differed from 
private sector or nonfederal pay.[Footnote 41] As shown in table 2, 
conclusions varied on which sector had the higher pay (which does not 
include benefits) and the size of pay disparities. All but one of the 
studies estimated the difference in pay after controlling for some 
personal and job-related attributes that can affect pay levels such as 
education and locality. This remaining difference is sometimes called 
the unexplained difference because it persists after controlling for 
attributes that can affect pay. 

However, the overall pay disparity number does not tell the whole 
story; each of the studies that examined whether differences in pay 
varied among categories of workers, found such variations (see table 
2). For example, CBO found that federal workers with graduate and 
professional degrees were paid less in comparison to the private 
sector, while workers without college degrees were paid more. 

Importantly, all of the study authors acknowledged that the data they 
used in their analyses had limitations which could affect their 
findings. Any comparison of the studies needs to take these data 
limitations into account. For example, studies that used the Census 
Bureau's Current Population Survey (CPS) were unable to directly 
control for years of work experience given this measure is not 
available in the CPS; 
some of the authors said that work experience is an attribute that 
affects how much a person is paid. Also, it was acknowledged that many 
federal jobs may not have equivalents in the private sector. 

Table 2: Selected Studies' Findings on Pay: 

Study authors and affiliations: Andrew Biggs and Jason Richwine--
American Enterprise Institute for Public Policy Research; 
Study title, date: Comparing Federal and Private Sector Compensation; 
June 2011; 
Study authors' findings on pay: 
* On average, federal workers' pay was higher than private sector 
workers' pay by an unexplained 14%; 
* Federal workers with a high school education received pay 22% higher 
than comparable private sector workers, while those with graduate 
degrees received 3.9% more on average. 

Study authors and affiliations: Congressional Budget Office; 
Study title, date: Comparing the Compensation of Federal and Private-
Sector Employees; January 2012; 
Study authors' findings on pay: 
* On average, federal workers' pay was higher than private sector 
workers' pay by an unexplained 2%; 
* Federal workers with a high school education earned pay about 21% 
higher than similar private sector workers on average. Federal workers 
with professional degrees (e.g., lawyers) or with doctorates earned 
pay about 23% lower than similar private sector workers. 

Study authors and affiliations: Chris Edwards--The Cato Institute; 
Study title, date: Federal Pay Continues Rapid Ascent; August 2009; 
Study authors' findings on pay: 
* On average, federal workers' pay was higher than private sector 
workers' pay by an absolute amount of 58%. Study did not estimate how 
much of this difference was explained by personal or job-related 
attributes (e.g., education, experience); 
* The difference was 25 percentage points higher than it was 8 years 
before. 

Study authors and affiliations: The President's Pay Agent; 
Study title, date: Report on Locality-Based Comparability Payments for 
the General Schedule, Annual Report of the President's Pay Agent 2010; 
March 2011; 
Study authors' findings on pay: 
* On average, federal workers' pay was lower than nonfederal (private, 
state, and local) workers' pay by an unexplained 24%; 
* The difference varied by locality pay area. 

Study authors and affiliations: The Project On Government Oversight 
(POGO); 
Study title, date: Bad Business: Billions of Taxpayer Dollars Wasted 
on Hiring Contractors; September 2011; 
Study authors' findings on pay: 
* On average, federal workers' pay was higher than private sector 
workers' pay by an unexplained 20% across the occupations studied[A]; 
* The direction of the differences varied across the occupations 
studied. For example, pay for federal Claims Assistants and Examiners 
was 24% lower than pay for similar private sector workers, while pay 
for federal Correctional Officers was 117% higher than pay for similar 
private sector workers. 

Study authors and affiliations: James Sherk--The Heritage Foundation; 
Study title, date: Inflated Federal Pay: How Americans Are Overtaxed 
to Overpay the Civil Service; July 2010; 
Study authors' findings on pay: 
* On average, federal workers' pay was higher than private sector 
workers' pay by an unexplained 22%; 
* This difference varied by occupation. For example, for lawyers and 
economists the unexplained difference was less than 10%, while for 
security guards and stock clerks it was close to 50%. 

Source: GAO analysis of selected studies. 

Notes: See appendix III for a complete description of the studies' 
details including attributes controlled for and statistical methods 
used. The six studies defined federal workers differently for the 
purpose of their analyses. The President's Pay Agent defined federal 
workers as those in GS and equivalent pay plans, POGO included federal 
workers in the GS and wage grade pay plans covering 35 selected 
occupations, and the remaining four studies included civilian 
employees except for U.S. Postal Service workers. 

[A] POGO presented the unexplained portion in total compensation (pay 
and benefits), not pay alone. However, the authors based the 
compensation difference on a pay difference, multiplied by similar 
factors for federal and private sector workers. Thus, according to the 
authors, the unexplained difference in pay would be similar to the 
unexplained difference in total compensation. 

[End of table] 

Selected Studies Used Three Basic Approaches to Analyzing Pay, with 
Varying Methodologies: 

The studies used three basic approaches to analyze differences in pay, 
as shown in table 3. Each author chose the approach they thought would 
best describe differences in pay. The Pay Agent is mandated by law to 
compare the rates of pay under the GS system with the rates of pay 
generally paid to nonfederal workers for the same levels of work 
within each pay locality, as determined on the basis of appropriate 
BLS surveys.[Footnote 42] 

Table 3: Approaches Used by Selected Studies in Analyzing Pay: 

Type of approach: Human capital; 
Description: Compares pay for individuals taking into account personal 
attributes (e.g., education, job experience) and other attributes 
(e.g., occupation, locality, firm size).[A] Demographic personal 
attributes such as race and gender are taken into account[B]; 

Studies that used the approach: 
* Biggs/Richwine; 
* CBO; 
* Sherk. 

Type of approach: Job-to-job; 
Description: Aligns and compares pay for similar jobs of various types 
based on job-related attributes such as occupation and level of work 
(e.g., entry-level, mid-level, senior level or finer distinctions by 
level). Does not take into account the personal attributes of the 
workers currently filling the jobs[A]; 
Studies that used the approach: 
* President's Pay Agent; 
* POGO. 

Type of approach: Trend analysis; 
Description: Illustrates broad trends in pay over time without 
controlling for attributes of the workers or jobs; 
Studies that used the approach: 
* Edwards. 

Source: GAO analysis of selected studies. 

[A] More information on attributes that can affect pay is provided in 
appendix III. 

[B] Authors using the human capital approach explained that it was 
important to account for the differences between the sectors in racial 
composition, gender, and other demographic characteristics so that the 
unexplained difference is isolated from these factors. 

[End of table] 

The studies' differing conclusions on the overall pay disparity 
between federal and private or nonfederal workers were affected by 
their basic approaches--human capital, job-to-job, and trend analysis. 
Across these approaches, data sources and types of attributes 
controlled for differed. Within each approach, conclusions differed 
due to studies' specific methodologies--specific attributes controlled 
for and statistical methods used. 

* Basic approaches: Across the three basic approaches, the differences 
in the data sources and types of attributes controlled for (personal 
or job-related) contributed to the differing conclusions.[Footnote 43] 

- Data sources: The type of approach the study authors chose 
influenced the data sources they used. Studies using the human capital 
approach used data from the CPS[Footnote 44] to determine the pay for 
federal and private sector workers. Studies using the job-to-job 
approach used data from BLS's National Compensation Survey (NCS) 
[Footnote 45] to determine pay for nonfederal (Pay Agent) and private 
sector (POGO) workers and data from OPM to determine pay for federal 
workers. For the trend analysis approach, Edwards used data from BEA's 
national income and product accounts (NIPA)[Footnote 46] tables to 
determine pay for federal and private sector workers. 

- Types of attributes: Most of the studies estimated the unexplained 
difference in pay, accounting for the fact that employees earn 
different amounts based on education, locality, and other personal and 
job-related attributes. However, studies using different basic 
approaches controlled for different types of attributes. Studies using 
the human capital approach controlled for attributes related to both 
the individual worker and the job the person occupied. Studies using 
the job-to-job approach controlled for only job-related attributes. 
The trend analysis approach did not control for attributes. 

* Specific methodologies: Within the human capital and job-to-job 
approaches, the studies controlled for different specific attributes 
and used different statistical methods, as shown in table 4. These 
differences led to differing conclusions. 

Table 4: Comparisons among the Selected Studies' Methodologies: 

Human capital approach: 

Similarities: 
* Controlled for both personal and job-related attributes including 
education, occupation, and locality; 
* Used a statistical method to determine the extent to which 
differences in pay are explained by different attributes; 
Key differences: 
* Biggs/Richwine and CBO controlled for firm size, while Sherk did not; 
* Biggs/Richwine used a regression and CBO and Sherk used a 
decomposition; 
* CBO's decomposition took into account the wider distribution of pay 
among private sector workers; 
How the differences affected conclusions: 
* The inclusion of firm size largely explains the differences between 
the findings of Biggs/Richwine and Sherk. According to Biggs/Richwine, 
if they had not controlled for firm size, they would have found almost 
the same pay disparity as Sherk; 

* According to CBO, its results largely differed from Biggs/Richwine 
because of the specific statistical method used which took into 
account the wider distribution of pay. Had CBO used the same method as 
Biggs/Richwine, its results for the overall pay difference would have 
been similar. 

Job-to-job approach: 

Similarities: 
* Controlled only for job-related attributes such as occupation; 
* Conclusions are only generalizable to the population analyzed; 
Key differences: 
* The Pay Agent matched pay rates for over 200 occupations, while POGO 
used 35 selected occupations that had been outsourced; 
* The Pay Agent matched pay rates on a locality pay area basis and 
POGO matched pay rates at a national level; 
* The Pay Agent controlled for level of work; 
How the differences affected conclusions: 
* The different sets of occupations analyzed and the fact that the Pay 
Agent controlled for level of work and matched pay rates by locality 
may have contributed to the different conclusions. 

Source: GAO analysis of selected studies. 

[End of table] 

Study Authors Differed on Which Approach Is Most Informative: 

The study authors and people with expertise in compensation issues 
that we interviewed differed in their views on which type of approach 
is most informative in comparing pay of workers across sectors. 
According to study authors who used the human capital approach, this 
approach is the standard method in the field of economics to compare 
workers' pay across sectors. The overall unexplained difference 
between federal and private sector pay is a way to measure the extent 
to which the federal government may be paying more or less for the 
services it receives from its workers relative to what those workers 
could earn in the private sector. These findings could help inform 
policy decisions regarding the pay of federal workers. However, study 
authors (including those who used the human capital model) and people 
with expertise in compensation issues did not suggest that the human 
capital approach be used for setting an individual's rate of pay. They 
explained that some of the personal attributes that are associated 
with analyzing differences in pay using a human capital approach are 
demographic in nature (e.g., race, gender) and not work-related. OPM 
officials added that they are not aware of any employers that use the 
human capital approach to set pay for their employees. 

The President's Pay Agent and POGO used the job to job approach in 
their analyses of pay differences, not the human capital approach. 
According to OPM officials who serve as staff to the President's Pay 
Agent, employees with the same human capital characteristics can 
choose to work in markedly different jobs with large variations in 
pay. POGO and some people with expertise in compensation issues said 
that the fundamental concept of setting pay based on the job, without 
taking account of the personal characteristics of individuals in 
similar jobs, is the most appropriate approach. They said it is not 
appropriate to pay individuals differently according to personal 
attributes, such as education or job experience, if they hold the same 
job. However, others said that matching individuals by occupation and 
level of work involved some subjective judgment and lacks 
transparency, which makes it difficult for other interested parties to 
understand the analysis. 

The President's Pay Agent has stated that it has serious concerns 
about a process that requires a single percentage adjustment in the 
pay of all white-collar civilian federal employees in each locality 
pay area without regard to the differing labor markets for major 
occupational groups, and it believes that reforms of the GS system 
should be considered. Specifically, the Pay Agent stated that the 
underlying model and methodology for estimating pay gaps should be 
reexamined to ensure that private sector and federal sector pay 
comparisons are as accurate as possible. 

Selected Studies Used Different Methods to Include Benefits in 
Estimates of Total Compensation: 

Five studies found a wide range of disparities in benefits as part of 
total compensation (pay and benefits) between the federal and private 
sector workforces, as shown in table 5. (The President's Pay Agent 
Report did not include an analysis of benefits as part of total 
compensation.)[Footnote 47] Most studies presented the disparity in 
terms of total compensation, not just the benefits portion, because 
the levels of some benefits--for example, most retirement benefits--
are a function of pay rates, years of service, and type of plan. The 
five studies included benefit comparisons in an effort to capture the 
cost of benefits to the federal government.[Footnote 48] 

As with their analyses of pay, the study authors acknowledged that 
limitations in data affected their analyses of total compensation and 
could affect their findings, as discussed below table 5. These 
limitations need to be taken into account when comparing the studies. 
Additionally, the studies do not all analyze the same group of federal 
workers; 
for example, POGO analyzed workers in 35 selected occupations. 

Table 5: Selected Studies' Findings on Total Compensation (Pay and 
Benefits): 

Study author: Biggs/Richwine; 
Study authors' findings: The average federal worker's total 
compensation package was 61% higher than that of a comparable private 
sector worker. 

Study author: CBO; 
Study authors' findings: The average federal worker's total 
compensation package was 16% higher than that of a comparable private 
sector worker. 

Study author: Edwards; 
Study authors' findings: The average federal worker's total 
compensation package was 100% higher than that of an average private 
sector worker. Study did not control for attributes in its analysis. 

Study author: POGO; 
Study authors' findings: On average, the federal worker's total 
compensation package was 20% higher than that of a private sector 
worker across the occupations studied. 

Study author: Sherk; 
Study authors' findings: The average federal worker's total 
compensation package was 30-40% higher than that of a comparable 
private sector worker. 

Source: GAO analysis of selected studies. 

[End of table] 

The wide range of estimates between the studies is due to the 
different data sources, types of benefits analyzed, and specific 
methodologies used. 

Data sources. Study authors agreed that available data were less 
adequate for comparing federal to private sector benefits than pay. 
Benefits data at the individual level are not available from a single 
source so the studies used multiple sources. This makes it challenging 
to compare across the sectors. For example, some data sources, such as 
the CPS, ask workers questions about their pay, but do not ask about 
the cost of their benefits because workers generally do not know the 
monetary value of their benefits. As a result, study authors used data 
sources such as the NCS that ask employers questions about the cost of 
their workers' benefits. Additionally, different studies drew from 
different data sources, contributing to the range of different results. 

* Biggs/Richwine used BLS's NCS data, specifically the Employer Costs 
for Employee Compensation portion for private sector worker data. For 
federal workers, they used the OPM/OMB civilian position full fringe 
benefit cost factor--a percent factor describing the cost of benefits 
relative to salaries.[Footnote 49] To capture benefits the OPM/OMB 
source did not cover, Biggs/Richwine used OPM's Federal Civilian Work 
Force Statistics: Work Years and Personnel Costs Report to determine 
paid leave, and the Annual Social and Economic Supplement of the CPS 
to estimate job security. 

* POGO used NCS data on private sector workers. For federal workers, 
it used the OPM/OMB civilian position full fringe benefit cost factor 
as Biggs/Richwine did. 

* CBO used more detailed data from the NCS and OPM for private sector 
and federal workers, respectively. These data were not publicly 
available. 

* Edwards and Sherk both used BEA's NIPA data. According to BEA, this 
data source includes annual intra-governmental payments to amortize 
the accumulated unfunded liability of the Civil Service Retirement and 
Disability trust fund. This reduces the data's accuracy for measuring 
compensation for current workers, according to the study authors that 
used the data.[Footnote 50] Sherk used OPM data to correct for this 
issue of federal retiree benefits. 

Benefits analyzed. The studies included different types of benefits in 
their analyses, contributing to the range of different results. In 
addition, the study authors made assumptions in determining the value 
of benefits. 

* All of the studies included health insurance, retirement benefits, 
and the employer portion of mandatory government benefits such as 
Social Security. 

* Biggs/Richwine, CBO, and POGO (for private sector workers only) 
included paid leave, while Sherk did not. 

* Biggs/Richwine included job security, asserting that federal workers 
are less likely to experience periods of unemployment than private 
sector workers and so can expect a higher income for a given salary 
over the course of a year. 

* All of the studies relied on estimates of future benefits, which 
requires assumptions to be made about the present value of the 
benefit, which may introduce uncertainty in the estimates. According 
to BEA, estimates of the present value of future benefits are 
inherently dependent on assumptions about the discount rate, 
participant separation rates, retirement ages, mortality, and even 
future pay increases and future inflation.[Footnote 51] As a result, 
the amount of money that has to be set aside today to pay for 
tomorrow's benefits could be different. 

Specific methodologies. It was not possible to estimate the cost of 
benefits directly while controlling for differences between the 
federal and private workforces, so most authors used various indirect 
methodologies. The indirectness increased uncertainty, and the wide 
range of methodologies led to different results. 

* CBO developed a model to estimate the relationship between federal 
workers' pay and the cost of the benefits they received, and an 
analogous model for private sector workers. CBO imputed employee 
benefits using those models, then compared benefits for federal and 
private sector workers controlling for personal and job-related 
attributes, just as they did for pay, to estimate the portion of the 
difference in total compensation unexplained by attributes. CBO was 
the only study to use a model that allowed for varying benefits-to-pay 
ratios for different pay levels. 

* Sherk calculated the difference in average total compensation for 
federal and private sector workers. He used his estimates of the 
unexplained difference in pay from the human capital model and applied 
this to the difference in average total compensation. He assumed the 
unexplained difference in total compensation was the same as the 
unexplained difference in pay. 

* Biggs/Richwine used different benefits-to-pay ratios for federal 
workers and private sector workers. They applied these ratios to the 
unexplained differences in pay from their human capital model to 
obtain the unexplained difference in total compensation. 
Biggs/Richwine assumed the unexplained difference in total 
compensation was the same as the unexplained difference in pay. 

* POGO used different benefits-to-pay ratios for federal workers and 
private sector workers. It applied these ratios to differences in pay 
for the selected occupations in each sector to obtain the percent 
difference in total compensation for these occupations. 

* Edwards calculated the difference in average total compensation for 
federal and private sector workers. He did not control for attributes 
between the workers. 

Concluding Observations: 

The findings of the selected studies comparing federal and private 
sector pay and total compensation varied because they used different 
approaches, methods, and data. When looking within and across the 
studies, it is important to understand these differences because they 
impact how the studies can be interpreted. On the one hand, the human 
capital approach compares pay for individuals taking into account 
personal attributes such as education and job experience. Study 
authors who used this approach said that analyzing federal and private 
sector workers' pay was a way to measure the extent to which the 
federal government may be overpaying or underpaying its employees 
compared to what they could earn in the private sector. On the other 
hand, the job-to-job approach compares pay for similar jobs on such 
job-related attributes as occupation and level of work rather than 
personal attributes. The President's Pay Agent, which used this 
approach, examined how pay for GS and nonfederal jobs compared for the 
same occupations and levels of work within the same locality pay areas 
with the goal of reducing existing pay disparities. Simply put, the 
differences among the selected studies are such that comparing their 
results to help inform pay decisions is potentially problematic. Given 
the different approaches of the selected studies, their findings 
should not be taken in isolation as the answer to how federal pay and 
total compensation compares with other sectors. 

As stated earlier, we have reported on the importance of considering 
the skills, knowledge, and performance of federal employees as well as 
the local labor market in making pay decisions.[Footnote 52] The 
President's Pay Agent has recommended that the underlying model and 
methodology for estimating the pay gaps be reexamined to ensure that 
private sector and federal sector pay comparisons are as accurate as 
possible. As a step in this direction, the administration recommended 
in its September 2011 deficit reduction proposal that Congress 
establish a Commission on Federal Public Service Reform composed of 
members of Congress, representatives from the President's Labor-
Management Council, members of the private sector, and academic 
experts to identify fundamental reforms for the federal government's 
human capital systems including compensation reform. As of June 2012, 
such a commission has not been established. 

Agency and Third-Party Comments: 

We provided a draft of this report to the Secretary of Commerce (for 
Census), the Commissioner of BLS, and the Directors of BEA and OPM for 
their review and comment. The Census Bureau had a technical comment on 
the draft report, which we incorporated into the final report. BEA and 
BLS had no comments on the draft report. OPM provided technical 
comments on the draft report, which we incorporated as appropriate. 

We provided applicable sections of the draft report to the authors of 
the selected compensation comparison studies for their review and 
comment. Biggs/Richwine and CBO provided technical comments, which we 
incorporated as appropriate. Edwards and Sherk did not have any 
comments on the draft section. POGO provided written comments (see 
appendix IV). In its letter, POGO stated it concurred with our draft 
finding that many factors hinder public and private sector pay 
comparisons, such as a lack of detailed data. POGO also suggested that 
we analyze OPM federal-nonfederal salary comparisons as part of our 
final report. We believe this information is already addressed in 
other sections of the report, which POGO did not receive for comment. 
In these sections, we discuss in detail how annual pay adjustments are 
determined including the President's Pay Agent process, which uses the 
comparisons referred to by POGO. 

As we agreed with your office, unless you publicly announce the 
contents of this report earlier, we plan no further distribution of it 
until 30 days from the date of this letter. At that time, we will send 
copies of this report to the appropriate congressional committees; 
the Secretary of Commerce; the Commissioner of BLS; the Directors of 
BEA, Census, and OPM; and other interested parties. In addition, the 
report will be available at no charge on the GAO website at 
[hyperlink, http://www.gao.gov]. 

If you or your staff have any questions about this report, please 
contact me at (202) 512-6806 or goldenkoffr@gao.gov. Contact points 
for our Offices of Congressional Relations and Public Affairs may be 
found on the last page of this report. GAO staff who made key 
contributions to this report are listed in appendix V. 

Signed by: 

Robert Goldenkoff: 
Director: 
Strategic Issues: 

[End of section] 

Appendix I: Objectives, Scope, and Methodology: 

This report examines (1) how annual pay adjustments for the General 
Schedule (GS) system are determined; (2) the extent to which the pay 
increases and awards available to GS employees recognize individual 
performance, and how the Office of Personnel Management (OPM) provides 
oversight of pay increases and awards; and (3) how selected studies 
compare federal and private sector pay and total compensation and the 
factors that may account for the different findings. 

To examine how the GS annual across-the-board and locality pay 
adjustments are determined, we reviewed legislation, OPM regulations, 
executive orders, Presidents' alternative pay plans, President's Pay 
Agent Reports, Federal Salary Council recommendations, OPM and Bureau 
of Labor Statistics (BLS) documents and reports, and reports by the 
Congressional Budget Office (CBO) and Congressional Research Service. 
[Footnote 53] We also examined how the methodology for determining 
locality pay has changed since the start of locality pay to the 
present. We interviewed selected members of the Federal Salary Council 
and its working group; the Council is to be made up of six 
representatives of federal employee groups and three experts in labor 
relations, and makes annual recommendations to the President's Pay 
Agent.[Footnote 54] We interviewed BLS officials, OPM officials who 
are knowledgeable about federal pay policy and serve as staff to the 
President's Pay Agent, and people with expertise in compensation 
issues including former federal officials experienced with pay and 
benefits issues. To provide background information illustrating a 
range of pay areas, we selected localities including the lowest paid 
locality, highest paid locality, and other localities to include a 
range of pay rates, population sizes, and geographic regions. 

To determine the extent to which pay increases and awards recognize 
individual performance, we analyzed legislation and OPM regulations on 
pay increases and awards available to employees in the GS pay system 
and identified those pay increases and awards that are determined in 
part by an individual's performance rating as measured by the agency's 
performance appraisal system. These pay increases and awards are: 
within-grade increases, quality step increases, and ratings-based cash 
awards. We recognize that there are other types of pay increases and 
awards that reflect an individual's contributions, such as suggestion/ 
invention and superior accomplishment awards, and pay increases that 
do not reflect an individual's performance at all including across-the-
board and locality pay adjustments. We identified eligibility 
requirements outlined in the legislation and regulations and clarified 
in OPM guidance that can affect a GS employee's eligibility for the 
increase or award, such as a waiting period given the individual's 
position in the pay grade, frequency of receiving an increase or 
award, and agency-specific criteria. 

To provide statistics on how the pay increases and awards were 
distributed among GS employees, we analyzed data from OPM's Central 
Personnel Data File (CPDF) for fiscal year 2011.[Footnote 55] The data 
we examined included only federal employees in the GS pay plan. The GS 
classification and pay system includes several pay plan codes: GS 
(covered by pay system established under 5 U.S.C. chapter 53, 
subchapter III); GM (covers employees covered by the Performance 
Management and Recognition System termination provisions of Pub. L. 
No. 103-89); GL (covers law enforcement officers who receive special 
base rates at grades 3-10 under section 403 of FEPCA); GP (covers GS 
physicians and dentists paid market pay under 38 U.S.C. § 7431(c)); 
and GR (covers physicians and dentists covered by the Performance 
Management and Recognition System termination provisions who are paid 
market pay under 38 U.S.C. § 7431(c)). In addition to the GS pay plan, 
the GM and GL pay plans are used in federal-nonfederal pay comparisons 
to set locality pay. For the purposes of this analysis, we excluded 
the GM and GL pay plans because the GS pay plan covers the majority of 
the individuals in the GS, GM, and GL pay plans. We also excluded the 
GP and GR pay plans since individuals in these pay plans are no longer 
limited to GS rates of pay and they receive market pay under a 
different pay system. 

We analyzed CPDF data for employees in the GS pay plan in the 
aggregate on the number, percentage and dollar amount of quality step 
increases, within-grade increases, and ratings-based cash awards; 
the amount of these increases and awards as a portion of the GS 
payroll (total adjusted basic pay for all employees in the GS pay 
plan); and the distribution of these increases and awards by rating 
pattern and rating levels. For the award/increase amounts as 
percentages of recipients' pay, we excluded employees whose adjusted 
basic pay amount was missing. For the calculations based on ratings, 
we excluded employees who were coded in CPDF as "not rated". The not 
rated code applies to an employee who has not yet received a rating of 
record under the agency performance appraisal system (e.g., someone 
newly hired). We also excluded employees whose ratings were missing 
due to data errors. For calculations based on rating levels or 
patterns (e.g., 5-level system), we excluded employees who were coded 
as not being covered by a performance appraisal system and generally 
do not have their performance appraised. We also excluded employees 
whose rating patterns were missing from the data due to data errors. 

To help determine the reliability and accuracy of the CPDF data 
elements used, we checked the data for reasonableness and the presence 
of any obvious or potential errors in accuracy and completeness. For 
example, we excluded employees who were coded as receiving an increase 
or award in error (e.g., individuals who received a level 1 or 2 
rating and a within-grade increase or ratings-based cash award) from 
our data. We also reviewed past GAO analyses of the reliability of 
CPDF data[Footnote 56] and interviewed OPM officials knowledgeable 
about the data to discuss the data's accuracy and steps OPM takes to 
ensure they are reliable. For example, in its checks of the data, OPM 
excludes data where the dollar value is zero for ratings-based cash 
awards and within-grade and quality step increases. Also, for within-
grade and quality step increases, OPM checks to make sure values for 
current and prior adjusted basic pay exist and the difference is 
greater than zero. On the basis of these procedures, we believe the 
data we used from the CPDF are sufficiently reliable for the purpose 
of this report.[Footnote 57] 

To describe how OPM provides oversight of pay increases and awards, we 
collected and analyzed OPM guidance to agencies on administering 
relevant pay increases and awards including regulations, memoranda, 
reports, fact sheets, and frequently asked questions. We interviewed 
OPM officials responsible for federal pay policies to discuss the 
implementation of the guidance and monitoring of agencies' use of 
increases and awards through reports and other means, and we 
interviewed OPM officials responsible for conducting human capital 
management evaluations at agencies on pay increases and awards to 
determine how they evaluate agencies' linkage of pay increases and 
awards with organizational results and monitor the overall GS system, 
among other things. 

To review selected studies that compare federal and private sector pay 
and total compensation and describe factors that help account for the 
different study findings, we reviewed the studies, summarized each 
study's methodologies and key findings, and confirmed the accuracy of 
our summaries with the authors. We compared and contrasted the 
differences between the approaches, methodologies, and data sources of 
the selected studies. We interviewed the selected study authors to 
obtain their views on the various methodologies and data sources 
available, why they chose the ones they used, and their conclusions 
based on their work. From July through December 2011, we conducted a 
detailed literature review of academic journals, agency and 
organization publications, and grey literature to identify the 
selected studies.[Footnote 58] We applied three criteria for study 
selection to the results--(1) studies that were published/issued since 
2005; (2) studies that include original analysis; and (3) studies that 
have the explicit and primary purpose of comparing federal and private 
sector pay and total compensation. Using these criteria, we identified 
at that time the following five studies as our proposed set to review 
(with the option to add other studies that may be issued during the 
course of our engagement and meet our criteria), see below: 

* Comparing Federal and Private Sector Compensation, Andrew Biggs and 
Jason Richwine, American Enterprise Institute for Public Policy 
Research, June 2011. (Co-author Richwine is from The Heritage 
Foundation.) 

* Federal Pay Continues Rapid Ascent, Chris Edwards, The Cato 
Institute, August 2009. 

* Report on Locality-Based Comparability Payments for the General 
Schedule, Annual Report of the President's Pay Agent 2010, The 
President's Pay Agent, March 2011. 

* Bad Business: Billions of Taxpayer Dollars Wasted on Hiring 
Contractors, The Project On Government Oversight, September 2011. 

* Inflated Federal Pay: How Americans Are Overtaxed to Overpay the 
Civil Service, James Sherk, The Heritage Foundation, July 2010. 

All of the selected studies except for the President's Pay Agent 
compared federal to private sector pay and total compensation. The 
President's Pay Agent compared federal to nonfederal pay (not 
benefits) and defined nonfederal as private sector, state government, 
and local government. We decided to include the President's Pay Agent 
Report as one of our selected studies given that it plays a major role 
in the overall discussion of federal pay comparability. The 
President's Pay Agent encompasses the Secretary of Labor and Directors 
of OPM and the Office of Management and Budget (OMB). To inform our 
understanding of the Pay Agent's report and process, we interviewed 
OPM officials who are staff to the Pay Agent, members of the Federal 
Salary Council and its working group including officials from the 
National Treasury Employees Union and the American Federation of 
Government Employees, and officials at BLS, which provides the 
nonfederal data used for the Pay Agent's analysis. 

Through our literature review, we also identified articles and papers 
that compare compensation in other sectors (state and local government 
to private sector, or industry to industry). Additionally, we 
identified discussions of the selected studies' findings and 
methodologies and of the issues of federal and private sector pay and 
total compensation comparison in general to further inform our review 
of the studies. We interviewed a number of individuals chosen for 
their expertise in compensation issues to obtain their views on the 
data sources for analyzing compensation and to provide a general 
context for the issues involved in comparing federal and private or 
nonfederal pay and total compensation. The findings regarding the 
selected studies are not based on input from these individuals. 
Representing a wide range of perspectives and experiences related to 
compensation issues, we identified these individuals through our 
literature review, background research on the topic, and 
recommendations from the study authors and other individuals 
knowledgeable about compensation issues. The selected individuals, 
some of whom were selected authors of the discussions noted above, 
included a university professor who has done research on compensation 
issues across sectors, a private sector compensation consultant, a 
staff member who researches compensation at an organization with a 
policy focus, and former senior federal officials who are experienced 
in federal pay and benefits issues. We interviewed officials from the 
Bureau of Economic Analysis (BEA), BLS, and Census Bureau to discuss 
how these agencies' data are used to measure federal and private or 
nonfederal pay, compensation, or benefits, and limitations of their 
data or surveys. We also interviewed officials from OPM involved in 
federal pay policies. We asked everyone we interviewed about their 
views on the strengths and limitations of the data sources used in the 
studies. 

We also asked everyone we interviewed, as applicable, to identify any 
additional studies that address our criteria for study selection. They 
did not identify any additional studies that met our criteria, but 
provided additional information, such as background articles. However, 
in January 2012, after our literature review was concluded, CBO issued 
a report: 

* Comparing the Compensation of Federal and Private-Sector Employees, 
Congressional Budget Office, January 2012. 

We included this study in our review because it met our criteria. This 
brought the total number of studies up to six. We interviewed the 
authors of the CBO study to obtain their views on the various 
methodologies and data sources available, why they chose the ones they 
used, their conclusions based on their work, and our understanding of 
their work. 

We did not examine the reliability or the appropriateness of the 
approaches, methods, and data used by the six selected studies in our 
scope, and we did not exclude any study on the basis of methodological 
quality. 

We conducted this performance audit from July 2011 to June 2012 in 
accordance with generally accepted government auditing standards. 
Those standards require that we plan and perform the audit to obtain 
sufficient, appropriate evidence to provide a reasonable basis for our 
findings and conclusions based on our audit objectives. We believe 
that the evidence obtained provides a reasonable basis for our 
findings and conclusions based on our audit objectives. 

[End of section] 

Appendix II: Implementation of Locality Pay: 

Locality Payments from 1994 to 2011: 

Even though the full locality payments recommended by the President's 
Pay Agent have not been provided after locality pay was implemented in 
1994, some locality increase has been provided each year since that 
time except during the pay freeze in 2011 and 2012. The President's 
Pay Agent reported that pay disparities were lower in 2011 than in 
1994 in 16 of the 21 pay localities that existed in both of those 
years. 

Figure 5 shows the relative pay rates for a GS-11 employee 
(approximately the midpoint grade level) in San Francisco and in the 
Rest of U.S. (the residual locality for areas not included in one of 
the other pay localities) and nonfederal equivalents based on the 
President's Pay Agent Reports. In 1994, the pay disparity between 
federal and nonfederal workers in San Francisco at the GS-11 level was 
30 percent, which decreased to 26 percent by 2011 (the most recent 
year for which disparity data is available). In 1994, the pay 
disparity between federal and nonfederal workers in the Rest of U.S. 
locality at the GS-11 level was 19 percent, which increased to 22 
percent by 2011.[Footnote 59] 

Figure 5: Annual Pay for a GS-11 Employee in San Francisco and Rest of 
U.S. Pay Localities and Reported Nonfederal Equivalents, 1994 to 2011: 

[Refer to PDF for image: multiple line graph] 

Annual pay: 
Year: 1994; 
GS-11, San Francisco: 41,119.3; 
Nonfederal equivalent, San Francisco: $53,290.1; 
GS-11, Rest of U.S. (non-metro areas): $39,603.1; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $47,163.9. 

Year: 1995; 
GS-11, San Francisco: $42,616.9; 
Nonfederal equivalent, San Francisco: $55,358.1; 
GS-11, Rest of U.S. (non-metro areas): $40,714.8; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $48,507.9. 

Year: 1996; 
GS-11, San Francisco: $42,848.9; 
Nonfederal equivalent, San Francisco: $58,139.7; 
GS-11, Rest of U.S. (non-metro areas): $41,890.5; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $50,574.8. 

Year: 1997; 
GS-11, San Francisco: $45,899.6; 
Nonfederal equivalent, San Francisco: $61,649.7; 
GS-11, Rest of U.S. (non-metro areas): $43,381.9; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $52,132.1. 

Year: 1998; 
GS-11, San Francisco: $47,447.3; 
Nonfederal equivalent, San Francisco: $64,072.7; 
GS-11, Rest of U.S. (non-metro areas): $44,732.9; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $54,077. 

Year: 1999; 
GS-11, San Francisco: $49,353; 
Nonfederal equivalent, San Francisco: $66,253.9; 
GS-11, Rest of U.S. (non-metro areas): $46,402.8; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $55,837.6. 

Year: 2000; 
GS-11, San Francisco: $52,358.3; 
Nonfederal equivalent, San Francisco: $68,959.2; 
GS-11, Rest of U.S. (non-metro areas): $48,611.6; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $58,133.9. 

Year: 2001; 
GS-11, San Francisco: $54,628.5; 
Nonfederal equivalent, San Francisco: $71,788.2; 
GS-11, Rest of U.S. (non-metro areas): $50,323.2; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $60,421.6. 

Year: 2002; 
GS-11, San Francisco: $57,592.7; 
Nonfederal equivalent, San Francisco: $75,174.4; 
GS-11, Rest of U.S. (non-metro areas): $52,413.4; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $62,753.6. 

Year: 2003; 
GS-11, San Francisco: $60,203.4; 
Nonfederal equivalent, San Francisco: $77,835.6; 
GS-11, Rest of U.S. (non-metro areas): $54,165.4; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $64,510.8. 

Year: 2004; 
GS-11, San Francisco: $63,201.8; 
Nonfederal equivalent, San Francisco: $80,658.9; 
GS-11, Rest of U.S. (non-metro areas): $56,184.2; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $66,370.7. 

Year: 2005; 
GS-11, San Francisco: $65,801.2; 
Nonfederal equivalent, San Francisco: $84,093.9; 
GS-11, Rest of U.S. (non-metro areas): $57,885.5; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $65,504.6. 

Year: 2006; 
GS-11, San Francisco: $68,055; 
Nonfederal equivalent, San Francisco: $85,845.5; 
GS-11, Rest of U.S. (non-metro areas): $59,352.1; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $67,670.4. 

Year: 2007; 
GS-11, San Francisco: $69,834.7; 
Nonfederal equivalent, San Francisco: $87,470.8; 
GS-11, Rest of U.S. (non-metro areas): $60,433.6; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $71,412.4. 

Year: 2008; 
GS-11, San Francisco: $73,089; 
Nonfederal equivalent, San Francisco: $88,434.5; 
GS-11, Rest of U.S. (non-metro areas): $62,266; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $74,459. 

Year: 2009; 
GS-11, San Francisco: $76,391.4; 
Nonfederal equivalent, San Francisco: $92,127.1; 
GS-11, Rest of U.S. (non-metro areas): $64,255.8; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $75,198.5. 

Year: 2010; 
GS-11, San Francisco: $77,722.1; 
Nonfederal equivalent, San Francisco: $94,184.5; 
GS-11, Rest of U.S. (non-metro areas): $65,204.8; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $76,990.2. 

Year: 2011; 
GS-11, San Francisco: $77,257.1; 
Nonfederal equivalent, San Francisco: $9,7009.7; 
GS-11, Rest of U.S. (non-metro areas): $64,999.3; 
Nonfederal equivalent, Rest of U.S. (non-metro areas): $79,570.2. 

Source: GAO analysis of OPM data. 

Note: There were 34 pay localities in the United States in 2011, 
composed of Alaska, Hawaii, 31 metropolitan areas, and the Rest of 
U.S. locality. San Francisco had the highest pay rate and Rest of U.S. 
had the lowest pay rate of these 34 areas. GS-11 salaries and 
nonfederal equivalents are based on the President's Pay Agent Reports. 

[End of figure] 

Changes to Surveys and Models for Calculating Locality Pay: 

There have been several changes to the surveys and models used for 
locality pay setting between the passage of the Federal Employees Pay 
Comparability Act (FEPCA) in 1990 and the Pay Agent process and report 
for 2011, the most current report available. Changes are illustrated 
in figure 6 and additional information is below. 

Figure 6: Surveys and Models for Calculating Locality Increases Have 
Changed Several Times from 1990 to 2011: 

[Refer to PDF for image: survey data] 

Law Overhauled System and Process (1990-1994): 

1990: Congress passed Federal Employees Pay Comparability Act (FEPCA). 
Established locality pay, specified that pay gap between federal and 
nonfederal workers should be reduced to 5% for each locality, as 
measured through Bureau of Labor Statistics (BLS) surveys. 

1991: BLS combined Area Wage Surveys and White-Collar Pay survey into 
the new Occupational Compensation Survey Program (OCSP), to provide 
data for the President's Pay Agent (PPA) to use in calculating 
locality pay. The new survey increased the number of private 
establishments surveyed, changed the set of occupations covered by the 
survey, and added state and local governments to the survey. 

1994: First locality-based comparability payments were provided. 

Major Change in Survey Data (1996-2005): 

1996: BLS stopped conducting OCSP, which used a fixed job list, and 
started conducting National Compensation Survey (NCS), which uses 
probability sampling of jobs, in order to reduce costs and respondent 
burden and expand occupational coverage. The Employment Cost Index 
(ECI) and a benefits survey were also merged into NCS. PPA began 
reviewing NCS, and used aged OCSP data during their review period.[A] 

1998: OPM/OMB/BLS working group was formed to improve NCS for use in 
setting pay.  

1999: OPM/OMB/BLS Working Group made recommendations to BLS that led 
to 5 improvements to NCS data. 
             
1. Improved Job Weighting: Develop a crosswalk between General 
Schedule (GS) jobs and the Standard Occupational Classification system 
to permit weighting data by federal employment. Implemented 2002 (in 
time for setting 2004 pay). 

2. Excluding Top Jobs: Identify and exclude survey jobs that would be 
above GS-15. Implemented 2002 (in time for setting 2004 pay). 

3. Econometric Salary Estimates: Develop an econometric model to 
estimate salaries for jobs not in the sample. Implemented 2002 (in 
time for setting 2004 pay).  

4. Improved Grading of Supervisory Jobs: Improve methods for grading 
supervisory jobs in sample. Implemented 2006 (in time for setting 2008 
pay).  

5. Job Grading System: Develop a four-factor job grading system with 
job family guides. Implemented 2006-2011 (in time for setting 2008-
2013 pay). 

2002-2005: PPA phased in use of NCS data in calculating locality pay. 

Integrating Occupational Employment Statistics Survey (2010 and after): 
                  
2010: BLS presented results of a regression method combining NCS and 
Occupational Employment Statistics (OES) data for calculating locality 
pay. OES has a larger sample size, covering more locality areas, and 
can be used for pay data by occupation and by area. NCS data can be 
used to specify GS level effects. BLS announced reductions in the size 
of NCS due to budget cuts, to take effect in 2011.  

2011: NCS budget cuts took effect. Federal Salary Council expressed 
concerns about volatility in the NCS/OES model for future years. PPA 
encouraged further review of the model, and said that providing more 
funding for NCS was not feasible before exploring other options. 

Source: analysis of legislation and PPA, Federal Salary Council, BLS, 
and OPM data. 

[A] See below for more information on the difference between the 
OCSP's fixed job list and the NCS's probability sampling approach. 

[End of figure] 

Major Change in Survey Data: Change from Fixed Job List to Probability 
Sampling: 

From 1991 to 1996, BLS conducted the Occupational Compensation Survey 
Program (OCSP) to collect data on pay of nonfederal workers.[Footnote 
60] OCSP used a fixed list of 3 to 8 positions in each of the five 
PATCO categories (Professional, Administrative, Technical, Clerical, 
and Other White-Collar) to represent the range of different white 
collar jobs. In 1996, there were 26 different positions - for example, 
Scientist (a professional position) and Key Entry Operator (a clerical 
position). Each position had one or more levels - for example, 
Scientist I to Scientist VIII; Key Entry Operator I and Key Entry 
Operator II. BLS referred to a particular position at a particular 
level (e.g., Scientist I) as a "job." BLS asked surveyed 
establishments to identify positions they had that corresponded to one 
of the representative jobs. BLS and OPM worked together to write, 
test, and maintain survey job descriptions tied to a single GS grade 
level. 

In 1996, BLS stopped conducting the OCSP and started conducting the 
National Compensation Survey (NCS), which uses probability sampling of 
jobs. BLS randomly selected positions at surveyed nonfederal 
establishments and determined which Standard Occupational 
Classification System job, PATCO category, and GS grade corresponded 
to the selected jobs. The Employment Cost Index (ECI) and a benefits 
survey were also merged into NCS. These changes were made to reduce 
costs and respondent burden and expand occupational coverage. 

The President's Pay Agent began reviewing the NCS in 1996, with input 
from the Federal Salary Council. During the time of their review, they 
used OCSP data, aged to a common reference date based on the ECI, to 
calculate pay disparities and recommend locality pay. In 1998, they 
determined that the NCS was not suitable for use without improvements, 
and a working group with representatives of OPM, BLS, and OMB was 
formed to recommend improvements. The working group made 
recommendations in 1999 that led to five improvements in the NCS data. 
The improvements were implemented starting in 2002, at which point the 
Pay Agent began to phase in use of NCS data. The recommendations are 
outlined in figure 6 above. 

Integrating Occupational Employment Statistics Survey and Reducing 
Sample Size of NCS: 

In 2008, the Federal Salary Council asked BLS to explore the use of 
additional sources of pay data so the Council could better evaluate 
the need for establishing additional locality pay areas, especially in 
areas where the NCS could not provide estimates of nonfederal pay. BLS 
developed a model to combine data from the Occupational Employment 
Statistics (OES) survey, another BLS survey, with NCS data in order to 
increase locality coverage. In 2010, due to budget cuts, BLS announced 
a reduction in the size of the NCS sample, and said that the model 
results from the combined surveys could still be used to calculate pay 
gaps. According to BLS officials, only the size of the NCS sample has 
changed, not the substance of what is collected, and the reduction 
should not affect the ability to determine levels of work. 

The Federal Salary Council wrote in its 2011 memo to the President's 
Pay Agent that it had concerns about the reduction. For 2011, the 
final year when the larger NCS data set was available, the Federal 
Salary Council reviewed modeled results both with and without the 
reduction, and found concerning discrepancies (about a 5 point average 
difference in computed pay gaps). In its memo, the Council recommended 
that the Pay Agent use only NCS data for setting pay until the new 
model is better understood, and that the full NCS survey be 
reinstated. The Council wrote that it plans to continue working with 
OPM and BLS to study the NCS/OES model. 

The President's Pay Agent wrote in its 2011 report dated March 2012 
that it does not consider more funding for NCS to be feasible before 
exploring other options. The Pay Agent supported the Council's plan to 
continue its review of the new model and to focus on the impact of 
dropping roughly half of the NCS sample on the volatility of the 
model. The Pay Agent also noted that the administration recommended 
Congress establish a Commission on Federal Public Service Reform 
composed of members of Congress, representatives from the President's 
Labor-Management Council, members of the private sector, and academic 
experts to identify fundamental reforms for the federal government's 
human capital systems including compensation reform. As of June 2012, 
such a commission has not been established. 

[End of section] 

Appendix III: Information on the Selected Studies' Data Sources and 
Methodologies for Analyzing Pay and Total Compensation: 

Pay Analysis: 

The six selected studies used different data sources and methodologies 
to analyze differences in pay between the federal and private sector 
or nonfederal workforces, as shown in table 6. They also varied 
slightly in how they defined the federal workforce and restricted 
their analysis of workers. 

Table 6: Selected Studies' Data and Methodologies Used in Analyzing 
Pay: 

Study authors and affiliations: Andrew Biggs and Jason Richwine--
American Enterprise Institute for Public Policy Research; 
Data source used: 
* Annual Social and Economic Supplement of the Current Population 
Survey (CPS) (2006-2010) for federal and private sector workers.[A]; 
Description of workforce: 
* Federal workers defined as civilian, non-postal workers; 
* Restricted the analysis to full-time, full-year workers earning more 
than $9,000; 
Attributes controlled for: 
* Hours worked per week; 
* Experience (age minus years of education minus 6); 
* Experience-squared to account for non-linear effects; 
* Years of education; 
* Firm size (6 categories); 
* Occupation (10 categories); 
* Immigration status; 
* Locality (state of residence); 
* Race; 
* Gender; 
* Marital status; 
* Year dummies to account for inflation; 
Also included interaction terms, such as 'experience × education' and 
'marital status × gender'; 
Methodology used: 
* Analyzed differences in annual pay (while controlling for hours 
worked); 
* Used a statistical method (log linear regression) to determine the 
extent to which differences in pay are explained by other attributes 
in addition to hours worked. 

Study authors and affiliations: Congressional Budget Office; 
Data source used: 
* Annual Social and Economic Supplement of the CPS (2006-2011) for 
federal and private sector workers; 
Description of workforce: 
* Federal workers defined as civilian, non-postal workers; 
* Restricted the analysis to full-time, full-year workers; 
Attributes controlled for: 
* Occupation (24 categories); 
* Education; 
* Experience (age minus years of education minus 6); 
* Age; 
* Firm size; 
* Locality (5 geographic regions and urban or rural location); 
* Sex; 
* Race; 
* Ethnicity; 
* Marital status; 
* Immigration status; 
* Citizenship; 
Methodology used: 
* Analyzed differences in hourly pay; 
* Used a statistical method (decomposition) to determine the extent to 
which differences in pay are explained by different attributes; 
* Decomposition took into account the wider distribution of pay among 
private sector workers. 
Study authors and affiliations: Chris Edwards--The Cato Institute; 
Data source used: 
* BEA's national income and product accounts (NIPA) data (2000-2008) 
for federal and private sector workers; 
Description of workforce: 
* Federal workers defined as civilian, non-postal workers; 
Attributes controlled for: None; 
Methodology used: 
* Calculated average percentage change of compensation across the 
federal and private sectors. 

Study authors and affiliations: The President's Pay Agent; 
Data source used: 
* OPM data (2010) for federal workers; 
* NCS data (2010) for nonfederal (private, state and local government) 
workers; 
Description of workforce: 
* Federal workers defined as GS and equivalent pay plan workers 
covering over 200 occupations; 
* Equivalent pay plans include GM and GL.[B]; 
Attributes controlled for: 
* Occupation (over 200 categories); 
* Level of work; 
* Locality (34 locality areas); 
Methodology used: 
* Matched pay rates in the federal government to pay rates in the 
nonfederal sector in over 200 federal occupations, 15 levels of work, 
and 34 locality areas; 
* Results of study apply to GS pay plan employees. 

Study authors and affiliations: The Project On Government Oversight 
(POGO); 
Data source used: 
* OPM data (2009) for federal workers; 
* NCS data (2009) for private sector workers; 
Description of workforce: 
* Federal workers defined as GS pay plan and wage grade pay plan (one 
of the blue-collar pay plans under the Federal Wage System) workers 
covering 35 selected occupations; 
Attributes controlled for: 
* Occupation (35 occupational categories); 
Methodology used: 
* Matched pay rates in the federal government with pay rates in the 
private sector at the national level by 35 occupation groups, covering 
500 service activities (such as Budget Analyst or Security Guard) that 
had been outsourced; 
* Results of study apply to the 35 selected occupations. 

Study authors and affiliations: James Sherk--The Heritage Foundation; 
Data source used: 
* Monthly CPS (2006-2009) for federal and private sector workers; 
Description of workforce: 
* Federal workers defined as civilian, non-postal workers; 
* Included only federal workers who work in public administration 
based on Census definitions.[C]; 
* Restricted the analysis to full-time workers between ages 25 and 65; 
* Excluded those earning less than $5 or more than $60 per hour; 
Attributes controlled for: 
* Age; 
* Education; 
* Marital status; 
* Race; 
* Gender; 
* Citizenship and nativity; 
* Locality (State); 
* Year of survey; 
* Size of metropolitan area; 
* Occupation (three specifications: no occupational controls, 22 broad 
occupational controls, 65 detailed occupations that exist in both the 
federal government and private sector); 
Methodology used: 
* Analyzed differences in hourly pay; 
* Used a statistical method (decomposition) to determine the extent to 
which differences in pay are explained by different attributes. 

Source: GAO analysis of selected studies. 

[A] The CPS collects data in two different ways. The monthly CPS is a 
monthly survey of households that are asked questions regarding pay 
and labor market status. The Annual Social and Economic Supplement is 
conducted once a year (February-April). This survey is sometimes 
called the March CPS. Biggs/Richwine and CBO used the Annual Social 
and Economic Supplement, while Sherk used the monthly CPS. Dates 
indicate when the surveys were conducted. The surveys ask about annual 
earnings over the prior calendar year. 

[B] See appendix I for more information on the other pay plans (GM and 
GL) that OPM considers to be equivalent to the GS pay plan in its 
comparisons between federal and nonfederal pay. 

[C] Census definitions of public administration are provided in the 
North American Industry Classification System code 92. Most federal 
workers are included, but postal workers and some other federal 
employees are excluded--for example, some employees of federal 
schools, hospitals, transportation facilities, utilities, and the 
Government Printing Office. 

[End of table] 

Controlling for Attributes in Analyzing Pay: 

Studies could control for many attributes--personal or job-related--to 
help explain the differences between federal and private sector pay, 
as shown in the previous table. The types of attributes the selected 
study authors controlled for depended on the type of approach used to 
analyze pay--human capital or job-to-job. For example, the human 
capital approach controls for personal attributes (e.g., education, 
job experience) and job-related attributes (e.g., occupation, firm 
size). The job-to-job approach involves controlling for job-related 
attributes (e.g., occupation, level of work) without considering the 
personal attributes of the workers. The trend analysis approach does 
not control for any attributes. Attributes such as occupation, level 
of work, firm size, locality, education, and job experience were 
considered relevant by several of the studies' authors and people with 
expertise in compensation issues that we interviewed. 

* Occupation:[Footnote 61] Controlling for occupation allows a study 
to account for different pay rates for different types of jobs. The 
distribution of occupations in the federal government is different 
from the private or nonfederal sector, which may be a factor that 
explains differences in pay. For example, according to the CBO study, 
33 percent of the federal workforce compared with 18 percent of the 
private sector workforce was in a professional occupation. A job-to-
job approach, as demonstrated by the study authors who used the 
approach, involves matching federal workers to equivalent positions in 
another sector. POGO limited its comparison to 35 selected 
occupations, while the Pay Agent used over 200 occupations. According 
to one of the people with expertise that we interviewed, one challenge 
with this approach is the difficulty of finding nonfederal equivalents 
for certain positions, such as the Federal Bureau of Investigation 
agents, that only exist in the government. Another person with 
expertise said matching occupations across sectors is a subjective 
process. In contrast, study authors using the human capital approach 
used fewer and much broader occupational groups. For example, 
Biggs/Richwine used 10 categories, while Sherk and CBO used 22 and 24, 
respectively. In addition to his analysis of the overall pay 
disparity, Sherk analyzed pay data with and without occupation 
controls, and reported that less was explained when occupation was not 
included in the analysis. 

* Level of Work: Controlling for level of work (or grade level) allows 
a study to account for different pay rates for different levels of job 
complexity and responsibility (e.g., entry-level, mid-level, senior 
level or finer distinctions by level). Level of work encompasses types 
of duties performed, the scope and effect of the work, the level of 
difficulty and responsibility, and the level of supervision received. 
It can be difficult to measure level of work since levels are defined 
differently in different workplace settings. Of the studies we 
examined, only the President's Pay Agent Report controls for level of 
work. For federal employment in the GS pay system, there are 15 grade 
levels. To compare these with levels in nonfederal workplaces, BLS 
economists ranked nonfederal positions based on four factors: 
knowledge, job controls and complexity, contacts (nature and purpose), 
and physical environment.[Footnote 62] One person with expertise 
observed that the human capital approach does not recognize that there 
are many different levels within an occupation such as accountant or 
lawyer. 

* Firm size: Controlling for firm size allows a study to account for 
the effect of the number of workers in a firm. Some of the study 
authors asserted that large firms tend to offer higher salaries and 
greater benefits than smaller firms, but they differed on the decision 
to control for this attribute. CBO and Biggs/Richwine felt that 
federal workers should be compared to private sector workers at 
similarly sized institutions (e.g., firms with at least 1,000 workers) 
and included a measure of firm size in their analyses. The reasons the 
authors cited included large firms requiring more occupational 
specialization or higher levels of skill than smaller firms. Sherk 
said he chose not to control for firm size because he views it as a 
proxy for individual ability in the private sector--the larger firms 
pay a premium to hire more capable individuals and the associated pay 
reflects that. He said this is not the case in the federal government; 
the federal government does not selectively hire employees from large 
corporations but competes for hiring with all sizes of firms in the 
private sector. Sherk felt that including firm size could bias results 
if more productive workers tend to work in larger firms in the 
private, but not the federal sector. A person with expertise we 
interviewed agreed that a larger firm would pay more and have better 
benefits and noted that large firms are in head-to-head hiring 
competition with the federal government. In 2008, the President's Pay 
Agent decided to include data from all establishments in its locality 
pay recommendations to increase the amount of data available for jobs. 
Since locality pay began in 1994, the Pay Agent had used only data 
from large establishments in its calculations.[Footnote 63] According 
to a person with expertise that we interviewed, the larger sample of 
data helps improve the quality of the job matching. 

* Locality: Controlling for locality differences allows a study to 
account for the pay rates in the local labor market. Differences in 
locality are central to a locality pay system and may be important in 
a nationwide comparison if federal workers tend to work in areas in 
which private sector or nonfederal pay is also high, such as certain 
urban areas. The Pay Agent analysis measures disparities for each of 
the 34 localities separately and within each locality matches federal 
workers to nonfederal worker equivalents.[Footnote 64] In cases where 
BLS did not survey a specific job at a specific level of work, it used 
a regression model to determine the salary that would be expected 
based on the worker's occupation, level of work, and locality. In 
contrast, the studies using the human capital approach explored 
national differences, and did not attempt to determine differences by 
locality areas. In their analysis, Biggs/Richwine and Sherk controlled 
for differences in locality by state while CBO used five geographic 
regions (Northeast, South, Midwest, West, and Washington, D.C. 
metropolitan area) and urban or rural location. 

* Education: Controlling for education allows a study to account for 
the effect of formal schooling that the workers have received. 
Increased education can increase productivity and result in higher pay 
by employers. The human capital approach can control for this 
attribute. Sherk and CBO controlled for education by grouping levels 
of education into degrees obtained as defined by the CPS such as high 
school, bachelor's degree, and other degrees as well as high school 
dropout and some college, no degree.[Footnote 65] In contrast, Biggs 
and Richwine controlled for education by converting degree obtained 
into a variable representing years of education. The POGO study did 
not control for education because according to the authors, education 
level does not reflect the individual's contribution to or output of 
the organization. An individual with a master's degree and a PhD may 
be paid the same pay rate in the market if they are producing the same 
output. 

* Job experience: Controlling for job experience allows a study to 
account for the length of time an individual has spent working. 
Experience both at a specific job and in general can affect pay, 
presumably because it can affect productivity, which can be accounted 
for in the human capital approach. Biggs/Richwine, CBO, and Sherk 
considered job experience a relevant attribute. However, the CPS does 
not include a direct measure of job experience. As a result, the 
studies use proxies to measure experience. For example, Biggs/Richwine 
and CBO used a common approach for measuring experience, "age minus 
years of education minus 6," while Sherk included age in his model. 
[Footnote 66] According to a university professor we interviewed who 
has done research on compensation issues across sectors, there is no 
data set that measures how long a private sector worker has been out 
of the workforce or how long a worker has been working for a given 
employer. Age can be used as a proxy, but age does not reflect time 
out of the workforce for child-rearing or other reasons. 

Total Compensation Analysis: 

The selected studies varied in the data sources used, benefits 
included, and methodologies chosen in analyzing benefits as a part of 
total compensation, as shown in table 7. The President's Pay Agent is 
mandated to analyze pay, not benefits, so its study is not included in 
the following table. 

Table 7: Selected Studies' Data and Methodologies Used in Analyzing 
Total Compensation (Pay and Benefits): 

Study author: Biggs/Richwine; 
Data sources used: Federal: 
* OPM/OMB cost factor on the cost of benefits relative to salaries--
civilian position full fringe benefit cost factor, as used in update 
to OMB circular A-76; 
* OPM data on levels of paid leave from Federal Civilian Work Force 
Statistics: Work Years and Personnel Costs Report, fiscal year 2005; 
* Annual Social and Economic Supplement of the CPS to estimate job 
security; 
Private: 
* BLS's NCS, Employer Costs for Employee Compensation for firms with 
greater than 100 employees; 
Benefits included: 
* Health insurance; 
* Life insurance; 
* Disability Insurance; 
* Retirement benefits--defined benefit and defined contribution plans; 
* Employer portion of mandatory government benefits such as Social 
Security and Medicare; 
* Paid leave; 
* Supplemental pay (e.g., overtime, shift differentials, nonproduction 
bonuses); 
* Benefit of greater job security (only federal); 
* Workers' compensation; 
* Unemployment programs; 
* Bonuses/awards (only federal); 
Methodology used: 
* Used estimates of benefits in the private sector from BLS data and 
estimates of benefits in the federal sector from OMB and OPM data; 
* Added additional compensation elements such as job security; 
* Assumed that the unexplained portion of the benefits as shown in the 
data was the same as the unexplained portion found in pay, estimated 
from a human capital model. This difference in benefits used took into 
account firm size, but not individual characteristics. 

Study author: CBO; 
Data sources used: Federal: 
* Unpublished data from OPM were used to obtain the relationship 
between pay and benefits; 
* Obtained information on attributes and pay of workers from the CPS; 
Private: 
* Unpublished data from BLS's NCS was used to obtain relationship 
between pay and benefits; 
* Obtained information on attributes and pay of workers from the CPS; 
Benefits included: 
* Health insurance; 
* Retirement benefits--defined benefit and defined contribution plans; 
* Employer portion of mandatory government benefits such as Social 
Security and Medicare; 
* Paid leave; 
* Workers' compensation; 
* Unemployment programs; 
Methodology used: 
* Obtained estimates of the relationship between pay and benefits for 
the private and federal sectors; 
* Using this relationship, for each individual in the CPS, used the 
rate of annual pay to obtain a predicted value of benefits. Then, 
applied a human capital model to estimate the portion of the 
difference in total compensation unexplained by attributes. 

Study author: Edwards; 
Data sources used: Federal and Private: 
* BEA's NIPA data: 
6.2D (Compensation of Employees by Industry); 
6.3D (Wage and Salary Accruals by Industry); 
6.5D (Full-Time Equivalent Employees by Industry); 
Benefits included: 
* Health insurance; 
* Life insurance; 
* Retirement benefits--defined benefit and defined contribution plans; 
* Employer portion of mandatory government benefits such as Social 
Security and Medicare; 
* Paid leave; 
Methodology used: 
* Calculated the difference in average total compensation between 
federal and private sector workers; 
* Did not determine the percentage explained by attributes. 

Study author: POGO; 
Data sources used: Federal: 
* OPM/OMB cost factor for the cost of benefits relative to salaries--
civilian position full fringe benefit cost factor, as used in update 
to OMB circular A-76; 
Private: 
* BLS's NCS cost factor for cost of benefits relative to salary; 
Benefits included: 
* Health insurance; 
* Life insurance; 
* Retirement benefits--defined benefit and defined contribution plans; 
* Employer portion of mandatory government benefits such as Social 
Security and Medicare; 
* Supplemental pay (only private); 
* Paid leave (only private); 
* Workers' compensation; 
* Unemployment programs; 
* Bonuses/awards (only federal); 
Methodology used: 
* Used the cost factor for benefits for federal (36.25%) and private 
(33.5%) sectors; 
* Applied these cost factors to pay estimates for the 35 occupations 
in each sector and determined the difference in total compensation. 

Study author: Sherk; 
Data sources used: Federal: 
* BEA's NIPA data: 
* 6.2D (Compensation of Employees by Industry); 
* 6.3D (Wage and Salary Accruals by Industry); 
* 6.5D (Full-Time Equivalent Employees by Industry); 
* OPM data on federal retirees who are receiving benefits under the 
Civil Service Retirement System; 
Private: 
* Annual Social and Economic Supplement of the CPS; 
* BEA's NIPA data; 
* 6.2D (Compensation of Employees by Industry); 
* 6.3D (Wage and Salary Accruals by Industry); 
* 6.5D (Full-Time Equivalent Employees by Industry); 
Benefits included: 
* Health insurance; 
* Life insurance; 
* Retirement benefits--defined benefit and defined contribution plans; 
* Employer portion of mandatory government benefits such as Social 
Security and Medicare; 
Methodology used: 
* Used the difference in average total compensation for federal and 
private sectors obtained from BEA data; 
* Used the CPS data to determine observable differences in pay between 
private sector and federal workers; 
* Assumed that the unexplained compensation premium in the BEA data 
was the same as the unexplained portion found in pay estimated from a 
human capital model; 
* Used OPM data to adjust for the fact that federal retirees who are 
receiving benefits under the Civil Service Retirement System are 
included in BEA's NIPA data for current federal workers. 

Source: GAO analysis of selected studies. 

[End of table] 

Data Sources Used in Compensation Analysis: 

The study authors had a variety of data sources to choose from in 
analyzing pay and total compensation. They chose the data sources for 
their studies based on their overall approach and data needs. The 
study authors and people with expertise in compensation issues that we 
interviewed identified strengths and limitations of two common data 
sources the studies used in analyzing pay or total compensation--the 
CPS and NCS. Agency officials who oversee these data sources also 
weighed in on the use of the data in analyzing compensation. 

* Current Population Survey. The CPS--and in particular, the monthly 
CPS--has a large sample size relative to other data sources enabling 
analyses that would not have been possible in data sets with a smaller 
sample size. According to Sherk, he used the monthly CPS because he 
needed at least 30 valid observations of occupations in both the 
public and private sectors for his analysis comparing detailed 
occupations. The Annual Social and Economic Supplement of the CPS has 
questions that are more in-depth than the monthly CPS and it contains 
measures of job tenure, educational degree, and firm size. Individuals 
interviewed for the monthly or Annual Social and Economic Supplement 
of the CPS are self-reporting in their responses, which can result in 
reporting errors. As an example of an error that could occur, 
individuals who work for a contractor employed by the federal 
government could identify themselves as federal employees, which would 
be incorrect. Census officials said that there are CPS interviewer 
manuals to assist interviewers in helping respondents with their 
answers. 

* National Compensation Survey. BLS conducts the NCS by interviewing 
employers, which allows for cost data on pay and benefits to be 
directly collected from employers as opposed to individuals self-
reporting the information. While the survey covers all sectors, it 
does not collect data on federal workers, which--according to the 
study authors who used the NCS--results in the need to piece together 
different sources of benefits information in order to get comparable 
data. The NCS also provides detailed pay information by occupational 
work level that is based on the duties and responsibilities of a job, 
which is a key source of information for the President's Pay Agent 
when determining locality pay adjustment amounts. Recently, the sample 
size for the NCS was reduced, and BLS has developed a model to 
determine locality pay using a combination of the NCS and the 
Occupational Employment Statistics (OES) survey. The OES is a larger 
survey with broader coverage of locality areas than the NCS, but it 
does not contain information on levels of work. (See appendix II for 
additional information on locality pay and the use of these surveys.) 

The following table provides additional details on the data sources 
relevant for analyzing compensation across sectors including a 
description of the data source and supporting methodology. 

Table 8: Data Sources Used in Analyzing Compensation: 

Data source, affiliated agency: Civilian Position Full Fringe Benefit 
Cost Factor; OPM and OMB; 
Description: OMB through Circular A-76 requires agencies to use 
standard cost factors to estimate certain costs of government 
performance including the cost of benefits for civilian personnel. 
This cost factor identifies the cost of benefits relative to salaries 
for federal workers; 
Covers the cost of benefits for federal civilian employees in four 
categories: 
* insurance and health benefit; 
* standard civilian retirement benefit; 
* Medicare benefit; and; 
* miscellaneous fringe benefits including workmen's compensation, 
bonuses, awards that are based on annual performance ratings, and 
unemployment programs; 
Does not include the value of paid leave, supplemental pay, and 
several small categories of benefits; 
Sample and methodology: 
* Cost factor based on OPM's actuarial analyses; 
* Updated most recently in an OMB memo from 2008. 

Data source, affiliated agency: Current Population Survey; 
Department of Commerce, Census Bureau; Department of Labor, BLS; 
Description: Provides labor statistics for the U.S. population; 
* The monthly CPS is a household survey with information on the 
employment and unemployment experience of the nation's population, 
classified by age, sex, race, education, geographic area, and a 
variety of other characteristics; 
* The Annual Social and Economic Supplement of the CPS adds questions 
to the monthly CPS covering household and family characteristics, 
marital status, geographic mobility, foreign-born population, income 
from the previous calendar year, poverty, work status/occupation, 
health insurance coverage, noncash benefits, and education; 
Sample and methodology: 
* The monthly CPS covers about 54,000 households. The sample is 
selected to represent the civilian population cutting across all 
sectors; 
* The Annual Social and Economic Supplement includes additional 
households. The sample includes estimates for the nation as a whole 
and for individual states and other geographic areas. This supplement 
is conducted annually (February-April). 

Data source, affiliated agency: Federal Civilian Work Force 
Statistics: Work Years and Personnel Costs Report; OPM; 
Description: Provides a wide range of federal workforce data on work 
years expended, payroll costs, fringe benefit expenses, types of leave 
used and its value, and other compensation items; Most recent 
published report is from fiscal year 2005; 
Sample and methodology: All executive branch agencies with over 100 
employees including the U.S. Postal Service. 

Data source, affiliated agency: Job Openings and Labor Turnover Survey; 
Department of Labor, BLS; 
Description: Provides data on job openings, hires, quits, layoffs and 
discharges, and other separations. These data serve as demand-side 
indicators of labor shortages at the national level. There is no 
demographic information on employees; Covers all U.S. nonagricultural 
industries in the public and private sectors; 
Sample and methodology: 
* Data are from a sample of approximately 16,000 U.S. businesses 
covering all nonagricultural industries in the public and private 
sectors; 
* Separation data are disaggregated into three types: voluntary 
(quits), involuntary (layoffs and discharges), and other (includes 
death and retirement); 
* BLS does not determine whether a separation is due to seasonal work. 

Data source, affiliated agency: National Compensation Survey; 
Department of Labor, BLS; 
Description: Covers local, regional, and national occupational 
earnings. Covers state and local governments and private industry; 
Components of NCS include: 
* Employment Cost Index: average quarterly changes in wage, benefit, 
and compensation rates for a fixed market basket of labor services; 
* Employer Costs for Employee Compensation: quarterly employer cost 
levels, including average hourly employer cost; 
* Employee Benefits Survey: incidence and provisions of employee 
benefits; 
Sample and methodology: 
* Conducted by interviews with employers covering 11,400 
establishments and 50,500 sampled occupations; 
* Federal employees are not included; 
* The North American Industry Classification System is used to 
stratify establishments by industry and the Standard Occupational 
Classification system is used to classify occupations. 

Data source, affiliated agency: National Income and Product Accounts; 
Department of Commerce, BEA; 
Description: Estimates total wages and total compensation including 
noncash compensation such as the employer's contribution for health 
insurance by industry across all sectors; 
Total compensation includes employer contributions to: social 
insurance--such as Social Security and Medicare; health insurance; 
pension and profit sharing plans. The value of paid leave is included 
in the total compensation calculations as wages and salaries; 
Sample and methodology: 
* Data largely originate from public sources, such as government 
surveys (e.g., the Quarterly Census of Employment and Wages) and 
administrative data (e.g., the Social Security Administration and OPM); 
* Data for general government compensation exclude federal workers in 
government enterprises such as postal workers. 

Data source, affiliated agency: Occupational Employment Statistics 
(OES) Survey; Department of Labor, BLS; 
Description: Covers occupational employment and wage rates of wage and 
salary workers in private industry, and state, local, and federal 
government employees on an hourly and annual basis, including mean and 
median earnings for all areas--national, state, Metropolitan 
Statistical Areas, and the nonmetropolitan balance of state areas; 

Sample and methodology: 
* 200,000 business establishments/6 months by substate area and 
industry, over 3 years, for 1.2 million total business establishments. 
Substate areas are Metropolitan Statistical Areas and one or more 
balance-of-state areas which cover the remaining non-Metropolitan 
Statistical Area portions of the state; 
* The North American Industry Classification System is used to 
stratify establishments by industry and the Standard Occupational 
Classification system is used to classify occupations. 

Data source, affiliated agency: Office of Personnel Management Federal 
Employee Data; OPM; 
Description: Provides selected data elements on federal civilian 
employees from the Central Personnel Data File and Enterprise Human 
Resources Integration-Statistical Data Mart. FedScope is the public 
access source for this data; 
Data includes age, gender, length of service, grade, occupation, 
salary, type of appointment, agency, location, and Metropolitan 
Statistical Area; 
Sample and methodology: 
* Data covers most agencies in the executive branch. Exceptions from 
coverage include the uniformed military, intelligence agencies, the 
White House, Tennessee Valley Authority, the Board of Governors of the 
Federal Reserve, and the U.S. Postal Service; 
* Coverage of the legislative branch is limited to the Government 
Printing Office, the U.S. Tax Court, and selected commissions. 

Source: GAO analysis of data sources for compensation. 

[End of table] 

[End of section] 

Appendix IV: Comments from the Project On Government Oversight: 

POGO: 
Project on Government Oversight
Exposing Corruption, Exploring Solutions: 
1100 G Street, NW, Suite 500: 
Washington, DC 20005: 
(p) 202-347-1122: 
pogo@pogo.org: 
[hyperlink, http://www.pogo.org] 

June 6, 2012: 

Mr. Robert Goldenkoff: 
Director, Strategic Issues: 
U.S. Government Accountability Office: 
441 G St, NW: 
Washington, DC 20548: 

Mr. Goldenkoff: 

Thank you for providing the Project On Government Oversight (POGO) 
with the opportunity to review the draft report. Federal Workers: 
Results of Studies on Federal Pay Varied Due to Differing     
Methodologies (GAO-12-564). 

Based on the draft report, or portion thereof, provided to POGO, we 
concur with the draft report's finding that many factors hinder public 
and private sector pay comparisons. For example, the lack of detailed 
data on both salaries and compensation (which includes salary, 
benefits, and all incidental costs that should be considered when 
comparing the public and private sector workforce) makes it nearly 
impossible to draw empirical conclusions about one sector being 
underpaid. Simply stated, federal agencies collect salary and 
compensation information pursuant to numerous models that permit 
agencies and others to manipulate data when conducting an analysis.  

The numerous formulas utilized in human capital and job-to-job 
comparisons often lead to very different results. Data collection and 
quality issues often hinder the analysis of information related to 
salary, benefits, education, locality, experience, and personal 
attributes, which inhibits a scientific review of public and private 
pay.  

We suggest that the final report and subsequent reports include an 
analysis of Office of Personnel Management (OPM) salary comparisons. 
Those comparisons utilize a consolidated human capital and job-to-job 
approach, which provides some evidence that public employees are 
underpaid when compared to the private sector.[Footnote 1] 
     
Finally, the draft report states that government and contractor cost 
comparisons were outside the scope of GAO's analysis. As we emphasized 
in our Bad Business report,[Footnote 2] this issue should not be 
ignored as the government considers ways to reduce federal government 
spending.  

Sincerely, 

Signed by: 

Scott H. Amey: 
General Counsel: 

Appendix IV Footnotes: 

1. One comparison clearly shows that federal of Personnel Management, 
"Use of Bureau    pay for a government engineer is lower than private 
sector pay. Office of Labor Statistics Data for Setting General 
Schedule  Pay." http://www.opm.gov/oca/pay/html/UsingBLSData.asp] 
(Downloaded June 6, 2012). 

2. Project On Government Oversight, Bad Business: Billions of Taxpayer 
Dollars Wasted on Hiring Contractors, September 13, 2012. [hyperlink, 
http://www.pogo.org/pogo-files/reports/contract-oversight/bad-
business/co-gp-20110913.html#Government%20Cost9/020Studies]. 

[End of section] 

Appendix V: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

Robert Goldenkoff, (202) 512-6806, goldenkoffr@gao.gov: 

Staff Acknowledgments: 

In addition to the contact named above, Trina Lewis (Assistant 
Director), Laurel Beedon, Benjamin Bolitzer, Sara Daleski, Karin 
Fangman, Robert Gebhart, Janice Latimer, Rebecca Shea, and Meredith 
Trauner made key contributions to this report. 

[End of section] 

Footnotes: 

[1] Numbers of federal civilian employees and personnel costs are 
based on the Office of Personnel Management's (OPM) Central Personnel 
Data File (CPDF). The CPDF data we analyzed include most executive 
branch civilian employees, and do not include the U.S. Postal Service, 
judicial branch employees, intelligence agencies, nor most legislative 
branch employees. Personnel costs include both pay and benefits for 
full-time, part-time, and temporary workers. OPM has transitioned from 
the CPDF to the Enterprise Human Resources Integration-Statistical 
Data Mart (EHRI-SDM) as of fiscal year 2010, but CPDF still exists as 
a quarterly extract from the EHRI-SDM. We used these quarterly 
extracts for our analysis. 

[2] Excluding the U.S. Postal Service. 

[3] GAO, Human Capital: Symposium on Designing and Managing Market-
Based and More Performance-Oriented Pay Systems, [hyperlink, 
http://www.gao.gov/products/GAO-05-832SP] (Washington, D.C.: July 27, 
2005). 

[4] Are Federal Workers Underpaid: Hearing Before the House 
Subcommittee On Federal Workforce, U.S. Postal Service and Labor 
Policy of the Committee on Oversight and Government Reform, 112TH 
Cong. (2011). 

[5] The President's Pay Agent Report addresses only pay. It does not 
analyze benefits. 5 U.S.C. § 5304(d)(1). 

[6] The CBO reports referred to here do not include the January 2012 
study we review in detail later in this report. 

[7] Federal Salary Council members are selected by the President. 

[8] We decided to include the President's Pay Agent Report as one of 
our selected studies given that it plays a major role in the overall 
discussion of federal pay comparability. The Pay Agent compares 
nonfederal (private sector, state government, and local government) 
and federal pay on an annual basis for locality pay setting. See 
appendix I for additional details on the selected studies. 

[9] Other pay systems include the Federal Wage System (also known as 
the Prevailing Rate System), which covers about 9 percent of federal 
workers in mostly blue collar positions; 
the Senior Executive Service pay system for senior level managers; 
and other systems that are specific to agencies, such as the federal 
financial regulatory agencies' pay systems. The costs and percentages 
presented here are based on OPM's CPDF. Total compensation includes 
both pay and benefits. 

[10] Pub. L. No. 429, 63 Stat. 954 (Oct. 28, 1949). 

[11] In 1967, Congress authorized the President to make adjustments in 
federal salary rates for 1968 and 1969. Section 212 of the Federal 
Salary Act of 1967, Pub. L. No. 90-206, title II, 81 Stat. 624, 634 
(Dec. 16, 1967). 

[12] This authority also applied to other pay systems, including the 
Foreign Service pay system. 

[13] GAO, Recruitment and Retention: Inadequate Federal Pay Cited as 
Primary Problem by Agency Officials, [hyperlink, 
http://www.gao.gov/products/GAO/GGD-90-117] (Washington, D.C.: Sept. 
11, 1990). 

[14] Pub. L. No. 101-509, title V, § 529, 104 Stat. 1389, 1427-1469 
(Nov. 5, 1990). GS employees permanently stationed in foreign 
countries do not receive locality pay. GS employees in Alaska, Hawaii, 
and U.S. territories and possessions began receiving locality pay in 
2010. Locality pay may be extended to certain categories of non-GS 
employees by the President's Pay Agent. 

[15] Other statutory pay systems such as the Foreign Service pay 
system provide the same across-the-board increase to employees as the 
GS pay system. 

[16] An exception was employees receiving higher special rates to 
address specific recruitment and retention problems. 

[17] [hyperlink, http://www.gao.gov/products/GAO/GGD-90-117]. 

[18] The Federal Salary Council has from time to time recommended 
criteria for selecting additional pay areas when BLS had funding for 
more surveys and also recommended criteria for including additional 
areas in existing locality pay areas. The Pay Agent makes the final 
decisions on pay areas. The current criteria to include a multi-county 
metropolitan area in an adjacent locality pay area are based on the 
number of GS employees and the level of commuting between the two 
areas. There were 28 pay localities (27 metropolitan areas plus the 
Rest of U.S.) in 1994 when the first locality payments were made. 
There have been several additions and subtractions since then. Under 
FEPCA, the establishment or modification of any pay locality boundary 
is to be achieved through regulations promulgated in accordance with 
notice and comment requirements. 5 U.S.C. § 5304(f)(2). 

[19] There were no across-the-board or locality pay rate increases 
provided for 2011 and 2012. See section 147 of the Continuing 
Appropriations Act, 2011, Pub. L. No. 111-242, 124 Stat. 2607 (Sept. 
30, 2010) as added by section 1(a)(2) of the Continuing Appropriations 
and Surface Transportation Extensions Act, 2011, Pub. L. No. 111-322, 
124 Stat. 3518 (Dec. 22, 2010). Federal employees continued to receive 
within-grade step increases as appropriate. 

[20] The ECI for September 2011 included a sample of about 12,600 
private sector establishments. 

[21] In evaluating economic conditions, the President is to consider a 
range of economic measures, including (but not limited to) Gross 
National Product, the unemployment rate, the budget deficit, and the 
Consumer Price Index. 

[22] We started our analysis with 1994, the first year in which the 
FEPCA formula, unmodified, was used. 

[23] In 1994, 2010, 2011, and 2012, the lower amount was the result of 
a law passed by Congress. In 1996 and 1998 the lower amount was due to 
the President's alternative pay plan. In 1995, a law passed by 
Congress provided for the same lower amount as provided for in the 
President's alternative pay plan. 

[24] See section 147 of the Continuing Appropriations Act, 2011, Pub. 
L. No. 111-242, 124 Stat. 2607 (Sept. 30, 2010) as added by section 
1(a)(2) of the Continuing Appropriations and Surface Transportation 
Extensions Act, 2011, Pub. L. No. 111-322, 124 Stat. 3518 (Dec. 22, 
2010). 

[25] 5 U.S.C. § 5304(e). In 2011, the employee organizations 
represented on the Federal Salary Council were the American Federation 
of Government Employees, the National Treasury Employees Union, the 
National Federation of Federal Employees, and the Fraternal Order of 
Police. 

[26] FEPCA specified that locality pay be phased in over 9 years. In 
1994, the gap between federal and nonfederal pay in each locality 
would be reduced by two-tenths of the amount needed to reduce the pay 
disparity to 5 percent; in 1995, by three-tenths of the amount needed; 
etc. 

[27] The recommended amount was provided in the first year and 25 
percent of the recommended amount was provided in the second year. 
Less than 15 percent of the recommended amount has been provided each 
year after that. 

[28] See section 147 of the Continuing Appropriations Act, 2011, Pub. 
L. No. 111-242, 124 Stat. 2607 (Sept. 30, 2010) as added by section 
1(a)(2) of the Continuing Appropriations and Surface Transportation 
Extensions Act, 2011, Pub. L. No. 111-322, 124 Stat. 3518 (Dec. 22, 
2010). 

[29] Agencies are authorized to make these type of awards pursuant to 
5 U.S.C. § 4503. OPM regulations provide that agencies may grant 
awards to employees or groups of employees in the form of cash, 
honorary or informal recognition, or time off. 5 C.F.R. § 451.104(a). 
The President is also authorized to make awards based on a suggestion, 
invention, or superior accomplishment or an exceptionally meritorious 
special act pursuant to 5 U.S.C. § 4504. 

[30] Agencies are required to provide within-grade increases to 
eligible employees under 5 U.S.C. § 5335. An agency decision to deny a 
within-grade increase is ultimately appealable by the employee to the 
Merit Systems Protection Board. 

[31] Agencies are permitted to provide ratings-based cash awards to 
eligible employees under 5 U.S.C. § 4505a. 

[32] Agencies are permitted to provide quality step increases to 
eligible employees under 5 U.S.C. § 5336. 

[33] Factors that are to affect GS employee eligibility for these pay 
increases and awards are specified in legislation and regulations and 
clarified in OPM guidance. 

[34] According to OPM regulations and guidance the five levels are 
defined as: level 1-unacceptable, level 2-level between fully 
successful and unacceptable, level 3-fully successful or equivalent, 
level 4-level between outstanding and fully successful, and level 5-
outstanding or equivalent. 5 C.F.R. § 430.208. OPM defines level 3 as 
"fully successful or equivalent" and level 5 as "outstanding or 
equivalent," allowing agencies to use different names for the same 
numerical performance ratings. For simplicity, we will use the terms 
"fully successful" and "outstanding" when referring to levels 3 and 5 
for the remainder of this analysis. 

[35] These data and other data throughout this section are based on 
our analysis of OPM's CPDF for employees in the GS pay plan for fiscal 
year 2011. For the analysis of employees' ratings throughout this 
section, we excluded employees who were coded in CPDF as "not rated." 
The not rated code applies to an employee who has not yet received a 
rating of record under the agency performance appraisal system (e.g., 
someone newly hired). We also excluded employees whose ratings were 
missing due to data errors. 

[36] In addition to 5-level systems (which cover 63 percent of GS 
employees), the following systems allow for distinctions between 
"fully successful" and higher levels of performance: systems with 
rating levels 1, 3, and 5 (covering 7 percent of GS employees); 
systems with levels 1, 3, and 4 (covering less than 1 percent of GS 
employees); systems with levels 1, 3, 4, and 5 (covering 5 percent of 
GS employees); systems with levels 1, 2, 3, and 5 (covering 6 percent 
of GS employees); and systems with levels 1, 2, 3, and 4 (covering 
less than 1 percent of GS employees). For data regarding rating level 
systems, we excluded employees who were coded as not being covered by 
a performance appraisal system and generally do not have their 
performance appraised. We also excluded employees whose rating 
patterns were missing from the data due to data errors. 

[37] The highest possible rating could have been "outstanding," "fully 
successful," or the level in between depending on the agency's 
appraisal system. 

[38] 5 U.S.C. § 4503 and § 4505a. While these authorities provide 
agencies with the authority to make awards, OPM approval is required 
for awards which exceed thresholds specified in statute or regulation. 

[39] OPM develops this report using information from CPDF and EHRI-
SDM. The report is distributed to agency awards administrators and 
performance management program managers and to others upon request. 

[40] Specifically, OPM officials determine, among other things, 
whether the justification for the award is adequately based on 
accomplishments, whether awards link to the specific accomplishments 
that support the agency's mission, and whether proper authorities have 
approved the award. 

[41] All of the selected studies except for the President's Pay Agent 
Report compared federal to private sector pay or compensation. The 
President's Pay Agent compared federal to nonfederal pay (not 
benefits) and defined federal workers as those in GS and equivalent 
pay plans and nonfederal workers as private sector, state government, 
and local government workers. The six studies defined federal workers 
differently for the purposes of their analyses. See appendix III for 
additional details on the studies' methodologies. 

[42] 5 U.S.C. § 5304(d)(1). 

[43] For a more detailed discussion about the data sources and 
methodologies used by the study authors, see appendix III. 

[44] The CPS collects data in two different ways. The monthly CPS is a 
monthly survey of households that are asked questions regarding pay 
and labor market status. The Annual Social and Economic Supplement is 
conducted once a year (February-April). Biggs/Richwine and CBO used 
the Annual Social and Economic Supplement, while Sherk used the 
monthly CPS. 

[45] The NCS is a survey of employee salaries, wages, and benefits at 
local, regional, and national levels. The NCS does not include federal 
workers. 

[46] The NIPA data tables estimate total wages and total compensation 
(including noncash compensation such as the employer's contribution 
for health insurance) by industry across federal and private sectors. 

[47] The President's Pay Agent Report addresses only pay. It does not 
analyze benefits. 5 U.S.C. § 5304(d)(1). 

[48] In its study, POGO also compared the total annual compensation 
for federal employees with federal contractor billing rates in order 
to determine whether the current costs of federal service contracting 
serves the public interest. Any comparison of government and 
contractor employee costs is outside the scope of our work. 

[49] OMB, through Circular A-76, requires agencies to use standard 
cost factors to estimate certain costs of government performance 
including the cost of benefits for civilian personnel. Based on 
actuarial analysis by OPM, this cost factor expresses the value of 
certain federal benefits as a percentage of salaries and is used to 
compare the compensation of federal workers relative to that of 
private sector workers who might perform the same duties. 

[50] BEA officials agree this amortization payment for the unfunded 
liability should not be counted as a benefit that accrues to current 
workers. In 2013, BEA is planning revisions to its time series of 
compensation estimates that will correct this problem. 

[51] More appropriate estimates of federal retirement costs may be 
available in the future. According to agency officials, BEA, which 
estimates the cost to the government of retirement benefits in order 
to derive compensation estimates for the NIPA tables, is moving to an 
accrual approach for both private and government compensation that 
will more accurately account for benefit promises to future retirees. 
As part of a comprehensive revision to the NIPA tables in 2013, this 
change will lead to revisions to BEA's time series of compensation 
estimates. 

[52] [hyperlink, http://www.gao.gov/products/GAO-05-832SP]. 

[53] The CBO reports referred to here do not include the January 2012 
study we review in detail later in this report. 

[54] Federal Salary Council members are selected by the President. 

[55] The CPDF is a database that contains individual records for most 
federal employees and is the primary governmentwide source for 
information on federal employees. OPM has transitioned from the CPDF 
to the Enterprise Human Resources Integration-Statistical Data Mart 
(EHRI-SDM) as of fiscal year 2010, but CPDF still exists as a 
quarterly extract from the EHRI-SDM. We used these quarterly extracts 
for our analysis. 

[56] For example, GAO, Women's Pay: Gender Pay Gap in the Federal 
Workforce Narrows as Differences in Occupation, Education, and 
Experience Diminish, [hyperlink, 
http://www.gao.gov/products/GAO-09-279] (Washington, D.C.: Mar. 17, 
2009). 

[57] We previously reported that governmentwide data from the CPDF 
were 96 percent or more accurate. See GAO, OPM's Central Personnel 
Data File: Data Appear Sufficiently Reliable to Meet Most Customer 
Needs, [hyperlink, http://www.gao.gov/products/GAO/GGD-98-199] 
(Washington, D.C.: Sept. 30, 1998). Also, in a document dated February 
28, 2008, an OPM official confirmed that OPM continues to follow the 
CPDF data quality standards and procedures contained in our 1998 
report. 

[58] Grey literature comprises documents produced by government, 
academia, business and industry in print and electronic formats that 
are protected by intellectual property rights, but not controlled by 
commercial publishers. 

[59] These percentages are the remaining pay gaps after taking 
locality pay into account. Relative pay rates for GS-11, the midpoint 
level, differ somewhat from relative pay rates for the average worker 
across grade levels. For example, while the disparity for GS-11 
workers in the Rest of U.S. locality went up by 3 percent between 1994 
and 2011, it went down by 2 percent for the average worker in the Rest 
of U.S. locality over that same time. 

[60] The OCSP and the National Compensation Survey (NCS), described 
later in this appendix, were designed to provide compensation data for 
use in implementing locality pay; at the same time, the surveys met a 
variety of other government needs as well as private sector needs. 

[61] Occupation is defined by BLS as a set of activities that 
employees are paid to perform. Employees that perform essentially the 
same tasks (such as lawyers or engineers) are in the same occupation, 
whether or not they work in the same industry. A number of jobs can be 
classified into an occupation. For example, a lawyer could practice 
family law or international treaty law. These different jobs would be 
classified under the occupation "lawyer." 

[62] BLS defines the fours factors and provides guidance to economists 
on how to apply the factors in leveling the jobs. For knowledge, BLS 
describes this factor separately for the broad categories of 
occupations, such as professional accounting, information technology, 
and business administration. "Knowledge" is defined as the level of 
knowledge expected for the occupation at progressively higher levels 
of work. For the other three factors, BLS provides descriptions that 
apply to all occupations. "Job controls and complexity" covers the 
amount and type of directions received, the complexity of the work, 
and the nature of the work within a job. "Contacts" covers the nature 
and purpose of contacts within a job but outside the supervisory chain; 
for example, contacts are routine and structured or nonroutine and 
unstructured, to convey simple information or to influence, control, 
or debate, with other employees of the same agency, other agencies, 
members of Congress, or others. "Physical environment" covers the 
nature of risk, and the amount of physical demands within the job. 

[63] A firm may consist of one or more establishments. 

[64] See appendix II for more information on the implementation of 
locality pay since 1994. 

[65] For analysis, CBO used 11 levels of educational attainment based 
on CPS education data. When reporting, CBO grouped these into five 
categories: high school diploma or less, some college, bachelor's 
degree, master's degree, and professional degree or doctorate. Sherk 
used 7 levels for analysis: high school dropout; high school graduate; 
some college, no degree; associate's degree; bachelor's degree; 
master's degree; and professional degree or doctorate. 

[66] Although measures of length of service are available for federal 
workers from OPM data, in the application of the human capital model, 
a common data source must be used for both sectors. This is because of 
the possibility of differences in how the different variables, such as 
education or pay, are measured across data sources. 

[End of section] 

GAO’s Mission: 

The Government Accountability Office, the audit, evaluation, and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the 
performance and accountability of the federal government for the 
American people. GAO examines the use of public funds; evaluates 
federal programs and policies; and provides analyses, recommendations, 
and other assistance to help Congress make informed oversight, policy, 
and funding decisions. GAO’s commitment to good government is 
reflected in its core values of accountability, integrity, and 
reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO’s website [hyperlink, http://www.gao.gov]. Each 
weekday afternoon, GAO posts on its website newly released reports, 
testimony, and correspondence. To have GAO e-mail you a list of newly 
posted products, go to [hyperlink, http://www.gao.gov] and select 
“E-mail Updates.” 

Order by Phone: 

The price of each GAO publication reflects GAO’s actual cost of 
production and distribution and depends on the number of pages in the 
publication and whether the publication is printed in color or black 
and white. Pricing and ordering information is posted on GAO’s 
website, [hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or 
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card, 
MasterCard, Visa, check, or money order. Call for additional 
information. 

Connect with GAO: 

Connect with GAO on facebook, flickr, twitter, and YouTube.
Subscribe to our RSS Feeds or E mail Updates. Listen to our Podcasts.
Visit GAO on the web at [hyperlink, http://www.gao.gov]. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 
Website: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]; 
E-mail: fraudnet@gao.gov; 
Automated answering system: (800) 424-5454 or (202) 512-7470. 

Congressional Relations: 

Katherine Siggerud, Managing Director, siggerudk@gao.gov, (202) 512-4400
U.S. Government Accountability Office, 441 G Street NW, Room 7125
Washington, DC 20548. 

Public Affairs: 
Chuck Young, Managing Director, youngc1@gao.gov, (202) 512-4800
U.S. Government Accountability Office, 441 G Street NW, Room 7149 
Washington, DC 20548.