2007 Nonforeign Area Cost-of-Living Allowance Survey Report: Pacific and Washington, DC, Areas, 74861-74896 [E8-28833]

Agencies

• Office of Personnel Management
• Personnel Management Office

[Federal Register Volume 73, Number 237 (Tuesday, December 9, 2008)]
[Notices]
[Pages 74861-74896]
From the Federal Register Online via the Government Printing Office [www.gpo.gov]
[FR Doc No: E8-28833]



Federal Register / Vol. 73, No. 237 / Tuesday, December 9, 2008 / 
Notices

[[Page 74861]]


-----------------------------------------------------------------------

OFFICE OF PERSONNEL MANAGEMENT


2007 Nonforeign Area Cost-of-Living Allowance Survey Report: 
Pacific and Washington, DC, Areas

AGENCY: U.S. Office of Personnel Management.

ACTION: Notice.

-----------------------------------------------------------------------

SUMMARY: This notice publishes the ``2007 Nonforeign Area Cost-of-
Living Allowance Survey Report: Pacific and Washington, DC, Areas.'' 
The Federal Government uses the results of surveys such as these to set 
cost-of-living allowance (COLA) rates for General Schedule, U.S. Postal 
Service, and certain other Federal employees in Alaska, Hawaii, Guam 
and the Northern Mariana Islands, Puerto Rico, and the U.S. Virgin 
Islands. This report contains the results of the COLA surveys conducted 
by the U.S. Office of Personnel Management in Hawaii, Guam, and the 
Washington, DC, area during the spring and summer of 2007.

DATES: Comments on this report must be received on or before February 
9, 2009.

ADDRESSES: Send or deliver comments to Charles D. Grimes III, Deputy 
Associate Director for Performance and Pay Systems, Strategic Human 
Resources Policy Division, U.S. Office of Personnel Management, Room 
7300B, 1900 E Street, NW., Washington, DC 20415-8200; fax: (202) 606-
4264; or e-mail: COLA@opm.gov.

FOR FURTHER INFORMATION CONTACT: J. Stanley Austin, (202) 606-2838; 
fax: (202) 606-4264; or e-mail: COLA@opm.gov.

SUPPLEMENTARY INFORMATION: Section 591.229 of title 5, Code of Federal 
Regulations, requires the Office of Personnel Management (OPM) to 
publish nonforeign area cost-of-living allowance (COLA) survey summary 
reports in the Federal Register. We are publishing the complete ``2007 
Nonforeign Area Cost-of-Living Allowance Survey Report: Pacific and 
Washington, DC, Areas'' with this notice. The report contains the 
results of the COLA surveys we conducted in Hawaii, Guam, and the 
Washington, DC, area during the spring and summer of 2007.

Survey Results

    Using an index scale with Washington, DC, area living costs equal 
to 100, we computed index values of relative prices in the Honolulu 
County, Hawaii County, Kauai County, Maui County, and Guam and the 
Commonwealth of the Northern Mariana Islands (CNMI) COLA areas. Then we 
added an adjustment factor of 5.0 to the Honolulu County price index, 
7.0 to the Hawaii County, Kauai County, and Maui County price indexes, 
and 9.0 to the Guam/CNMI price index and rounded the results to the 
nearest whole percentage point. The results indicate a reduction in the 
COLA rates for all Pacific areas.

Office of Personnel Management.
Michael W. Hager,
Acting Director.

2007 Nonforeign Area Cost-of-Living Allowance Survey Report: Pacific 
and Washington, DC, Areas

Table of Contents

Executive Summary

1. Introduction
    1.1 Report Objectives
2. Preparing for the Survey
    2.1 COLA Advisory Committees
    2.2 Pre-Survey Meetings
    2.3 Survey Item Selection
    2.3.1 Special Considerations
    2.4 Outlet Selection
    2.5 Geographic Coverage
3. Conducting the Survey
    3.1 Pricing Period
    3.2 Non-Housing Price Data Collection
    3.2.1 Data Collection Teams
    3.2.2 Data Collection Process
    3.3 Housing (Rental) Price Data Collection
4. Analyzing the Results
    4.1 Data Review
    4.2 Special Price Computations
    4.2.1 K-12 Private Education
    4.2.2 Health Insurance
    4.2.3 Water Utilities
    4.2.4 Energy Utilities Model
    4.2.5 Rental Data Hedonic Models
    4.3 Averaging Prices by Item and Area
    4.4 Computing Price Indexes
    4.4.1 Geometric Means
    4.4.2 Special Private Education Computations
    4.5 Applying Consumer Expenditure Weights
5. Final Results
6. Post Survey Meetings

List of Appendices

Appendix 1: Prior Survey Results: 1990-2006
Appendix 2: Estimated DC Area Middle Income Annual Consumer 
Expenditures
Appendix 3: COLA Survey Items and Descriptions
Appendix 4: COLA Rental Survey Data Collection Elements
Appendix 5: Utility Usage and Calculations
Appendix 6: Hedonic Rental Data Equations and Results
Appendix 7: Final Living-Cost Results for COLA Areas

Executive Summary

    The Government pays cost-of-living allowances (COLAs) to Federal 
employees in nonforeign areas in consideration of living costs 
significantly higher than those in the Washington, DC area. The Office 
of Personnel Management (OPM) conducts living-cost surveys to set the 
COLA rates. The methodology for conducting these surveys is prescribed 
in regulation at subpart B of part 591 of title 5 of the Code of 
Federal Regulations.
    This report provides the results of the COLA surveys OPM conducted 
in the spring and summer of 2007 in Honolulu County, Hawaii County, 
Kauai County, Maui County, Guam, and the Washington, DC area. The 
report details our comparison of living costs in the Pacific areas with 
living costs in the Washington, DC area.
    For the surveys, we contacted about 1,300 outlets and collected 
approximately 5,500 prices on more than 240 items representing typical 
consumer purchases. We also collected about 2,800 prices on rental 
housing. We then combined the data using consumer expenditure 
information from the Bureau of Labor Statistics. The final results are 
living-cost indexes, shown in Table 1. These indexes compare living 
costs in the Pacific COLA areas to those in the Washington, DC area. 
The index for the DC area (not shown) is 100.00 because it is, by law, 
the reference area. The living-cost indexes shown in Table 1 include 
the adjustment factor prescribed at 5 CFR 591.227.

              Table 1--Final Living-Cost Comparison Indexes
------------------------------------------------------------------------
                        Allowance area                            Index
------------------------------------------------------------------------
Honolulu County, HI...........................................    121.37
Hawaii County, HI.............................................    111.71
Kauai County, HI..............................................    118.14
Maui County, HI...............................................    123.62
Guam/CNMI.....................................................    119.98
------------------------------------------------------------------------

1. Introduction

1.1 Report Objectives

    This report provides the results of the 2007 Pacific nonforeign 
area cost-of-living allowance (COLA) survey conducted by the U.S. 
Office of Personnel Management (OPM) in the spring and summer of 2007. 
In addition to providing these results, the report describes how we 
prepared for and conducted the survey, and how we analyzed the results. 
The results show comparative living-cost differences between the 
Pacific areas, i.e., Honolulu County, Hawaii County, Kauai County, Maui 
County, and Guam, and the Washington, DC area. By law, Washington, DC 
is the base or ``reference'' area for the COLA program.

[[Page 74862]]

2. Preparing for the Survey

2.1 COLA Advisory Committees

    Before conducting the Pacific survey, OPM established COLA Advisory 
Committees (CACs) in Honolulu, the Hawaii County areas of Hilo and 
Kailua Kona, Kauai, Maui, and Guam. The settlement of Caraballo, et al. 
v. United States, No. 1997-0027 (D.V.I.), August 17, 2000, provides for 
employee involvement in the administration of the COLA program. As in 
previous surveys, we found it valuable to involve employee and agency 
representatives in planning and conducting the surveys and in reviewing 
the survey results.
    Each CAC is composed of approximately 12 agency and employee 
representatives from the survey area and 2 OPM representatives. The 
functions of the CACs include the following:

--Advising and assisting OPM in planning COLA surveys;
--Providing or arranging for data collection observers during COLA 
surveys;
--Advising and assisting OPM in reviewing survey data;
--Advising OPM on its COLA program administration, including survey 
methodology;
--Assisting OPM in disseminating information to affected employees 
about the surveys and the COLA program; and
--Advising OPM on special situations or conditions, such as hurricanes 
and earthquakes, as they relate to OPM's authority to conduct interim 
surveys or implement some other change in response to conditions caused 
by a natural disaster or similar emergency.

2.2 Pre-Survey Meetings

    To help OPM prepare for the COLA surveys, the CACs held 3-day 
meetings in Honolulu, Hilo, Kailua Kona, Kauai, Maui, and Guam. The 
CACs reviewed the preliminary outlet and item lists for the surveys. 
The committee members researched the outlets and availability and 
appropriateness of the items in each area and made recommendations 
concerning the survey. We incorporated these recommendations into the 
survey design.
    We found the work of the CACs to be extremely helpful and 
informative. The CACs' knowledge of the local area, the popularity of 
items and outlets, and other information about the COLA area were 
invaluable in helping plan the survey.

2.3 Survey Item Selection

    As described in Sections 2.1 and 2.2, we consulted with the CACs as 
we selected survey items. We identified items to reflect a wide array 
of items consumers typically purchase. To determine what consumers 
purchase, we used the Bureau of Labor Statistics (BLS) 2002/2003 
Consumer Expenditure Survey (CES). We aggregated CES expenditures into 
the following nine major expenditure groups (MEGs):

--Food,
--Shelter and Utilities,
--Household Furnishings and Supplies,
--Apparel,
--Transportation,
--Medical,
--Recreation,
--Education and Communication, and
--Miscellaneous.

    We further subdivided each MEG into primary expenditure groups 
(PEGs). In all, there were 45 PEGs. For example, we subdivided Food 
into the following nine PEGs:

--Cereals and Bakery Products;
--Meats, Poultry, Fish, and Eggs;
--Dairy Products;
--Fresh Fruits and Vegetables;
--Processed Foods;
--Other Food at Home;
--Nonalcoholic Beverages;
--Food Away from Home; and
--Alcoholic Beverages.

    To select survey items, we chose a sufficient number of items to 
represent each PEG and reduce overall price index variability. To do 
this, we applied the following guidelines:
    Each survey item should be--

--Relatively important (i.e., represent a fairly large expenditure) 
within the PEG;
--Relatively easy to find in both COLA and DC areas;
--Relatively common, i.e., what people typically buy;
--Relatively stable over time, e.g., not a fad item; and
--Subject to similar supply and demand functions.

    In all, we selected over 240 non-housing items to survey. Appendix 
2 shows how we organized the CES data into MEGs and PEGs, identifies 
the Detailed Expenditure Categories (DECs) for which we chose survey 
items, and shows estimated DC area middle income annual consumer 
expenditures for each DEC and higher level of aggregations.
    Appendix 3 lists the non-housing items we surveyed and their 
descriptions. Each of these items is specifically described with an 
exact brand, model, type, and size whenever practical. Thus, we priced 
exactly the same items or the same quality and quantity of items in 
both the COLA and DC areas. For example, OPM priced a 10.75-ounce can 
of Campbell's Chicken Noodle Soup in both the COLA and DC areas because 
it is typical of canned soups, and consumers commonly purchase it.
2.3.1 Special Considerations
    Health Insurance: It was not practical to compare the prices of 
exactly the same quality and quantity of health insurance between the 
COLA and Washington, DC, areas because the same array of plans is not 
offered in each area, and a significant proportion of Federal employees 
in both the COLA and DC areas subscribe to plans not available 
nationwide. To compare the employee health benefits premiums of these 
often highly different plans, OPM would have to adjust for differences 
in benefits and coverage. Research conducted by the parties prior to 
the Caraballo settlement indicated this would not be feasible.
    Therefore, we use the non-Postal Service employee's share of the 
Federal Employees Health Benefits premiums by plan for each plan 
offered in each area. OPM maintains these data in the Central Personnel 
Data File (CPDF), including the number of white-collar Federal 
employees enrolled in each plan. As described in Section 4.2.3 below, 
we used these data to compute the average ``price'' of health insurance 
for Federal employees in the COLA and DC areas.
    Housing: For housing items, we survey rental rates for specific 
kinds or classes of housing and collect detailed information about each 
housing unit. We survey the following classes of housing:

--Four bedroom, single family unit, not to exceed 3200 square feet;
--Three bedroom, single family unit, not to exceed 2600 square feet;
--Two bedroom, single family unit, not to exceed 2200 square feet;
--Three bedroom, single family unit, not to exceed 2,600 square feet;
--Two bedroom, single family unit, not to exceed 2,200 square feet;
--Three bedroom apartment unit, not to exceed 2,000 square feet;
--Two bedroom apartment unit, not to exceed 1,800 square feet; and
--One bedroom apartment unit, not to exceed 1,400 square feet.

    For each housing unit we surveyed, we assessed approximately 80 
characteristics about the unit. For example, we determined the number 
of bedrooms, bathrooms, square footage, and whether there was a garage, 
air conditioning, security systems, and recreational facilities. 
Appendix 4 lists the types of detailed information we collected. We did 
not collect homeowner data, such as mortgage payments, maintenance 
expenses, or

[[Page 74863]]

insurance. Under the Caraballo settlement, the parties agreed to adopt 
a rental equivalence approach similar to the one BLS uses for the 
Consumer Price Index. Rental equivalence compares the shelter value 
(rental value) of owned homes, rather than total owner costs, because 
the latter are influenced by the investment value of the home (i.e., 
influenced by what homeowners hope to realize as a profit when they 
sell their homes). As a rule, living-cost surveys do not compare how 
consumers invest their money.
    We survey rents and use that as a surrogate for rental equivalence. 
In late 2004 and 2005, we conducted special research, the General 
Population Rental Equivalence Survey (GPRES), to obtain additional rent 
and rental equivalence information. The goal was to determine whether 
we should adjust the rent index before using it to estimate homeowner 
rental values. The analyses showed no adjustments should be made. 
Therefore, use of the rents to estimate rental equivalence is 
appropriate. We published the GPRES results in a Federal Register 
notice on July 31, 2006, at 71 FR 43228.
    Although we surveyed rental rates for the same classes of housing 
in each area, the type, style, size, quality, and other characteristics 
of each unit varied within each area and between the COLA and DC areas. 
As described in Section 4.2.5, we used hedonic regression analyses to 
hold these characteristics constant between the COLA and Washington, 
DC, area to make rental price comparisons.

2.4 Outlet Selection

    Just as it is important to select commonly-purchased items and 
survey the same items in both the DC area and COLA areas, it is 
important to select outlets frequented by consumers and find comparable 
outlets in both the COLA and DC areas. To identify comparable outlets, 
we categorize outlets by type (e.g., grocery store, convenience store, 
discount store, hardware store, auto dealer, and catalog outlet) and 
then survey only specific items at each outlet type. For example, we 
survey grocery items at supermarkets in all areas because most people 
purchase their groceries at such stores and because supermarkets exist 
in nearly all areas. Selecting comparable outlets is particularly 
important because significant price variations may occur between 
dissimilar outlets (e.g., comparing the price of milk at a supermarket 
with the price of milk at a convenience store).
    We used the above classification criteria and existing data 
sources, including previous COLA surveys, phone books, and various 
business listings, to develop initial outlet lists for the survey. We 
provided these lists to the CACs and consulted with them on outlet 
selection. The committees helped us refine the outlet lists and 
identify other/additional outlets where local consumers generally 
purchase the survey items.
    We also priced some items by catalog, and when we did, we priced 
the same items by catalog in the COLA areas and in the DC area. We 
priced 9 items by catalog in the Pacific areas. All catalog prices 
included any charges for shipping and handling and all applicable 
taxes, including excise taxes.
    In all, we surveyed prices from approximately 1,300 outlets. In the 
COLA survey areas, we attempted to survey three popular outlets of each 
type, to the extent practical. For some outlet types, such as local 
phone service, there were not three outlets. In some areas, there was 
not a sufficient number of businesses to find three outlets of each 
particular type. In the Washington, DC, area, we surveyed up to nine 
popular outlets of each type, three in each of the DC survey areas 
described in Table 2.

2.5 Geographic Coverage

    Table 2 shows the Pacific COLA and DC survey area boundaries. We 
collected non-housing prices in outlets throughout the Pacific areas 
described in Table 2. To collect rental housing data, we contracted 
with Delta-21 Resources, Incorporated, a research organization with 
expertise in housing and rental data collection. Delta-21 surveyed 
rental rates in locations within these areas.
    In selecting the locations and sample sizes within each location, 
we used 2000 census data showing the relative number of Federal 
employees and housing units by zip code. We allocated the rental sample 
objectively, requiring Delta-21 to attempt to obtain more rental 
observations in locations with a relatively large number of Federal 
employees and housing units and fewer observations in locations with a 
relatively small number of Federal employees and housing units. 
Although the process provided a rational way to allocate the sample, 
Delta-21 was limited ultimately by how many units were available for 
rent within a location. Under the contract, Delta-21 surveyed only 
units available for rent. It did not survey all renter-occupied 
housing.

                Table 2--Survey and Data Collection Areas
------------------------------------------------------------------------
   COLA areas and reference
            areas                             Survey area
------------------------------------------------------------------------
Honolulu County..............  City and County of Honolulu.
Hawaii County................  Hilo area, Kailua Kona/Waimea area.
Kauai County.................  Kauai Island
Maui County..................  Maui Island.
Guam/CNMI....................  Guam.
Washington, DC-DC............  District of Columbia.
Washington, DC-MD............  Montgomery County and Prince Georges
                                County.
Washington, DC-VA............  Arlington County, Fairfax County, Prince
                                William County, City of Alexandria, City
                                of Fairfax, City of Falls Church, City
                                of Manassas, and City of Manassas Park.
------------------------------------------------------------------------
Note: For selected items, such as golf and air travel, these survey
  areas include additional geographic locations beyond these
  jurisdictions.

    To collect non-housing data in the DC area, we divide the area into 
three survey areas, as shown in Table 2. We collect non-housing prices 
in outlets throughout these areas. We survey certain items, such as 
golf, in areas beyond the counties and cities specified in Table 2. We 
also survey the cost of air travel from Ronald Reagan Washington 
National Airport, Washington Dulles International Airport, and 
Baltimore/Washington International Airport (BWI) and survey the price 
of a 5-mile taxi ride originating at these airports. Both Dulles and 
BWI are outside the counties and cities shown in Table 2. Nevertheless, 
DC area residents commonly use both airports.

[[Page 74864]]

    Delta-21 surveyed rental housing rates throughout the DC area. We 
do not divide the DC area into three separate survey areas for rental 
housing data collection but rather treat the area as a single survey 
area. As with the Pacific COLA areas, we used Census data to select 
specific locations and sample sizes within the DC area. Delta-21 
collected data accordingly within these locations.

3. Conducting the Survey

3.1 Pricing Period

    We collected data from early March through May 2007. We collected 
non-housing price data concurrently in the Pacific areas in March and 
collected the bulk of the DC area data in April and May. Delta-21 
collected rental data sequentially in the DC area, Guam, Kauai, Kailua 
Kona/Waimea, Hilo, Maui, and Honolulu County from March through July 
2007.

3.2 Non-Housing Price Data Collection

3.2.1 Data Collection Teams
    In both the COLA and Washington, DC, areas, OPM central office 
staff collected non-housing price data. In the COLA areas, data 
collection observers designated by the local COLA Advisory Committees 
accompanied the OPM data collectors. The data collection observers 
advised and assisted the data collectors in contacting outlets, 
matching items, and selecting substitutes. The observers also advised 
OPM on other living-cost and compensation issues relating to their 
areas. We found the observers to be a valuable resource in conducting 
the local area surveys.
    Because of logistical considerations, cost, and the fact OPM 
central office staff is very knowledgeable about the DC area, we did 
not use COLA Advisory Committee data collection observers in the 
Washington, DC, area. However, we made all of the DC area data 
available to the COLA Advisory Committees. This included both housing 
and non-housing data. The non-housing data showed the individual prices 
by item, store, and survey location as well as averages. The housing 
data included a photograph and a rough sketch of the layout of the 
rental unit. We also provided the COLA Advisory Committees with maps 
showing where each rental unit is located.
3.2.2 Data Collection Process
    The data collector/observer teams obtained most of the data by 
visiting stores, auto dealers, and other outlets. The teams also priced 
some items, such as bank interest, piano lessons, and private education 
tuition, by telephone. As noted in Section 2.4, we surveyed some items 
via catalog, including all shipping costs and any applicable taxes in 
the price. We also collected other data, such as sales tax rates and 
airline fares, from Web sites on the Internet.
    For all items subject to sales and/or excise taxes, we added the 
appropriate amount of tax to the price for computing COLA rates. We 
added 4.712 percent in Honolulu County and 4.166 percent in Hawaii 
County, Kauai County, and Maui County to account for the Hawaii general 
excise tax on businesses. In the DC area, sales tax rates varied by 
city, and some sales tax rates also varied by item, such as restaurant 
meals, within a location. Guam currently has no general sales or 
business tax that is passed on to the consumer separately at the time 
of sale.
    The data collectors collected the price of the item at the time of 
the visit to the outlet. Therefore, with certain exceptions, the data 
collectors collected the sale price if the item was on sale, and we 
used that sale price in the COLA calculations. The exceptions include 
coupon prices, going-out-of-business prices, clearance prices, mail-in 
rebates, and area-wide distress sales, which we do not use. We also do 
not collect automobile ``sale'' or negotiated prices. Instead, we 
obtain the sticker (i.e., non-negotiated) price for the model and 
specified options. The prices are the manufacturer's suggested retail 
price (including options), destination charges, additional shipping 
charges, appropriate dealer-added items or options, dealer mark-up, and 
taxes, including sales tax and licensing and title fees.

3.3 Housing (Rental) Price Data Collection

    As noted in Section 2.5, we contracted for the collection of rental 
housing data with Delta-21, which collected data in the Pacific areas 
and in the DC area. We arranged for COLA Advisory Committee observers 
to accompany Delta-21 rental data collectors for a limited period 
during the local rental surveys. The rental data collected included 
rental prices, comprehensive information about the size and type of 
dwelling, number and types of rooms, and other important amenities that 
might influence the rental price. Appendix 4 lists the data elements 
Delta-21 collected.
    Delta-21 identified units for rent from various sources, including 
rental property managers, realtor brokers, listing services, newspaper 
ads, grocery store bulletin boards, and drive-by observation. Delta-21 
then visited each rental unit, took a photograph of the unit, made a 
sketch of the floor plan based on exterior dimensions and shape, and 
noted the unit's longitude and latitude coordinates. We used longitude 
and latitude to (1) determine the distance of the rental unit from 
major commercial and Government centers, (2) to correlate census tract 
data (e.g., median income) for the tract in which the unit was located, 
and (3) to map each unit's location. As discussed in Section 4.2.5, we 
used certain census tract data elements along with the data Delta-21 
collected to determine the relative price of rents. OPM made the rental 
data available to the COLA Advisory Committees, including the 
photographs, sketches, and maps.

4. Analyzing the Results

4.1 Data Review

    During and after the data collection process, the data collectors 
reviewed the data for errors and omissions. This involved reviewing the 
data item-by-item and comparing prices across outlets within an area to 
spot data entry errors, mismatches, and other mistakes.
    After all of the data had been collected in both the COLA areas and 
the Washington, DC, area, we again reviewed the data by item across all 
of the areas. One purpose was to spot errors not previously detected, 
but the principal reason was to look at substitute items.
    A substitute is an item similar to but not exactly the same as the 
specified survey item. For example, we may specify a 32-ounce bottle of 
Heinz Ketchup as one of the items to survey. However, during the survey 
we may discover some allowance area stores do not carry this item, but 
all carry a 24-ounce bottle of Hunt's Ketchup. Therefore, we will price 
the 24-ounce Hunt's Ketchup in the allowance areas and in the DC area 
as a substitute. We will use the substitute price information in place 
of the price of the originally specified item.

4.2 Special Price Computations

    After completing the data review, we made special price 
computations for five survey items: K-12 private education, Federal 
Employees Health Benefits premiums, water utilities, energy utility 
prices, and rental housing prices. For each of these, we used special 
processes to calculate appropriate values for each survey area.
4.2.1 K-12 Private Education
    One of the items we surveyed is the average annual tuition for 
private education, grades K-12. As in previous surveys, we found 
tuition rates varied by grade level. Therefore, we computed

[[Page 74865]]

an overall average tuition ``price'' for each school surveyed by 
averaging the tuition rates grade-by-grade. Section 4.4.2 below 
describes the additional special use factor we applied to the average 
tuition rates in the price comparison process.
4.2.2 Health Insurance
    As noted in Section 2.3.1, we surveyed the non-Postal employees' 
premium for the various Federal Employees Health Benefits (FEHB) plans 
offered in each survey area. Using enrollment information from the 
CPDF, we computed two weighted average premium costs--one for self-only 
coverage and another for family coverage--for white-collar Federal 
employees in each of the COLA areas and in the Washington, DC, area. As 
shown in Table 3, we then computed an overall weighted average premium 
for each survey area by applying the number of white-collar Federal 
employees nationwide enrolled in self-only and family plans. We used 
the overall weighted average premiums as ``prices'' in the price 
averaging process described in Section 4.3.

                      Table 3--2007 Average FEHB Premiums for Full-Time Permanent Employees
                                          [Non-Postal Employees' Share]
----------------------------------------------------------------------------------------------------------------
                                                                                         Bi-weekly      Annual
                                                                  Self        Family      weighted     weighted
                          Location                              premium      premium      average      average
                                                                                          premium      premium
----------------------------------------------------------------------------------------------------------------
Honolulu County.............................................       $36.22       $80.14       $62.72    $1,636.32
Hawaii County...............................................       $35.48       $79.13        61.82     1,612.84
Kauai County................................................       $35.34       $80.53        62.61     1,633.45
Maui County.................................................       $36.30       $80.60        63.03     1,644.41
Guam/CNMI...................................................       $39.77      $102.42        77.57     2,023.75
DC Area.....................................................       $45.20       $93.96        79.93     2,085.32
Nationwide Enrollment.......................................      615,389      936,075
Enrollment Percentage.......................................       39.67%       60.33%
----------------------------------------------------------------------------------------------------------------

4.2.3 Water Utilities
    OPM surveyed water utility rates in each of the COLA and 
Washington, DC, survey areas. To compute the ``price'' of water 
utilities, OPM assumed the average monthly water consumption in each 
area was 7,600 gallons. This is consistent with the consumption amount 
OPM used in the previous COLA survey. OPM used this quantity along with 
the rates charged to compute the average monthly water utility cost by 
survey area. OPM used these average monthly costs as ``prices'' in the 
price averaging process described in Section 4.3 below.
4.2.4 Energy Utilities Model
    For energy utilities (i.e., electricity, gas, and oil), OPM 
collected from local utility companies and suppliers in the COLA and DC 
survey areas the price of various energy utilities used for lighting, 
cooking, cooling, and other household needs over a 12-month period. OPM 
then used the results of a heating and cooling engineering model to 
determine how many kilowatt hours of electricity, cubic feet of gas, 
and/or gallons of fuel oil are needed in each area to maintain a 
specific model home at a constant ambient temperature of 69 degrees 
when heat is used or 72 degrees when cooling is used. The engineering 
model uses local home construction information and climatic data from 
the National Oceanic and Atmospheric Administration and also includes 
the amount of electricity needed to run standard household appliances 
and lighting. For each survey area, OPM calculated the cost to heat and 
cool the model home using the different heating fuels and electricity 
for lighting and appliances. Although some homes use additional energy 
sources, such as wood, coal, kerosene, and solar energy, OPM did not 
price or include these in the calculations because, based on the 
results of the 2000 census, relatively few homes use these as primary 
energy sources.
    For the Pacific areas, OPM surveyed the price of electricity to 
compute home energy costs because the 2000 census indicated electricity 
is the primary energy source in more than 95 percent of the homes in 
Hawaii and Guam. In the DC area, OPM surveyed the costs of all three 
fuels (gas, oil, and electricity) and applied all taxes, fees, and fuel 
cost adjustments in effect for the 12-month period. OPM used 
percentages based on the usage of the different fuels in each survey 
area to compute a weighted average utility fuel cost for the area. 
Appendix 5 shows the energy requirements, relative usage percentages, 
and total costs by area. OPM used these total costs as the ``price'' of 
utilities in the COLA rate calculations.
4.2.5 Rental Data Hedonic Models
    As discussed in Sections 2.5 and 3.3, OPM hired a contractor to 
collect rental data, including rents and the characteristics of each 
rental unit. As described in Section 3.3, we collated the rental data 
with census tract information published by the Bureau of the Census 
using the longitude and latitude of the rental properties. We used 
census tracts, which are relatively small geographically, as surrogates 
for neighborhoods. We believe the census tract characteristics, such as 
the percentage of school age children, reflect the character and 
quality of the neighborhoods in which the rental units are found.
    OPM uses hedonic regression analysis, which is a type of multiple 
linear regression analysis, to compare rents in the COLA areas with 
rents in the DC area. Multiple linear regression is a type of 
statistical analysis used to determine how the dependent variable (in 
this case rent) is influenced by the independent variables (in this 
case the characteristics of the neighborhood and rental unit). In 
regression analyses, it is very important to choose the independent 
variables with great care, making certain only those meeting certain 
statistically significant thresholds are used in the analysis. To 
select the independent variables, OPM uses a special procedure 
developed jointly by OPM and the Technical Advisory Committee. (The 
Technical Advisory Committee was established under the Caraballo 
settlement and is composed of three economists with expertise in 
living-cost comparisons.) We call this procedure the Variable Selection 
Protocol (VSP).
    VSP is a multi-step procedure that uses objective criteria to 
eliminate independent variables with little

[[Page 74866]]

statistical significance in the regression. It also removes variables 
with inexplicable signs and variables that negatively affect the 
precision of the rent indexes. An example of an inexplicable sign is 
when the landlord provides an amenity (e.g., a microwave), and the 
variable has a negative sign. In essence, this is the same as saying on 
average when the landlord did not provide a microwave, the property 
rented for more than when the landlord did provide a microwave.
    How VSP drops variables that negatively affect the precision of 
rent indexes is a bit more complicated to explain. The key variable in 
the regression is the survey area, i.e., Honolulu, Hawaii County, Maui, 
Kauai, Guam, and the Washington, DC, area. As with all variables in the 
regression, these variables have parameter estimates; but the survey 
area parameter estimates are especially important because they become 
the rent indexes for each of the survey areas. Therefore, it is 
important that the survey area parameter estimates be as accurate as 
practicable. The accuracy is measured by the standard error of the 
survey area parameter estimate. In the last steps of VSP, the protocol 
tests each of the variables in the model and drops variables that if 
retained would raise the standard errors of the survey area parameter 
estimates.
    Using VSP, we selected variables with the greatest statistical 
significance. The variables are listed below and are shown in the 
regression output in Appendix 6.
    --Age of unit;
    --Age of unit squared;
    --Number of bathrooms;
    --Number of bedrooms;
    --Unit type (detached house, row/townhouse, high rise apartment, 
garden apartment, and other (in-home apartments, duplex/triplex/
quadplex units and other));
    --Number of square feet combined (i.e., ``crossed'') with unit 
type;
    --Square footage squared;
    --Neighborhood condition (above average, average, or below 
average);
    --Full kitchen (variable values range from 0-1 with three possible 
levels: 0, .5, or 1--variable receives .5 if unit has a refrigerator 
and .5 if it has a range or oven);
    --Electricity (landlord provides electricity);
    --Furniture (landlord provides furniture);
    --Percent BA index (percentage of population in the census tract 
with a baccalaureate degree or higher level of education divided by the 
percentage of the population in the survey area with a baccalaureate 
degree or higher level of education);
    --Percent school age index (percentage of population in the census 
tract of school age divided by the percentage of the population in the 
survey area of school age);
    --Percent below the poverty level index (percentage of population 
in the census tract with income below the poverty level divided by the 
percentage of the population in the survey area with income below the 
poverty level);
    --Survey year (2006 or 2007 DC area only); and
    --Survey area (Honolulu, Hawaii County, Maui, Kauai, Guam, or the 
DC area).
    We included the survey year variable in the regression calculations 
because, based on the recommendation of the Technical Advisory 
Committee, we use two years of DC area rental data. We find adding data 
from the previous year significantly reduces the standard error of the 
survey area parameter estimate.
    As is common in this type of analysis and as was done in the 
research leading to the Caraballo settlement, OPM uses semi-logarithmic 
regressions. As noted previously in this section, the regression 
produces parameter estimates for each independent variable, including 
survey area. When the regression uses the Washington, DC, area as the 
base, the regression produces parameter estimates for each of the COLA 
survey areas: Honolulu, Hawaii County, Maui, Kauai, Guam. The exponent 
of the survey area parameter estimate (i.e., after the estimate is 
converted from natural logarithms) multiplied by 100 (following the 
convention used to express indexes) is the survey area's rent index. 
This index reflects the difference in rents in each of the COLA survey 
areas relative to the Washington, DC, area, while holding constant 
important neighborhood and rental unit characteristics captured in the 
survey and census data.
    OPM makes a technical adjustment in the above calculations to 
correct for a slight bias caused by the use of logarithms because the 
exponent of the average of the logarithms of a series of numbers is 
always less than the average of the numbers. Therefore, we added one-
half of the standard deviation of the survey area parameter estimate 
before converting from natural logarithms. (See Arthur Goldberger, 
``Best Linear Unbiased Prediction in the Generalized Linear Regression 
Model,'' Journal of the American Statistical Association, 1962.) Table 
4 shows the resulting rent indexes. We used these indexes as ``prices'' 
in the price averaging process described in Section 4.3.

                          Table 4--Rent Indexes
------------------------------------------------------------------------
                            Area                              Rent index
------------------------------------------------------------------------
Honolulu County............................................       115.89
Hilo Area..................................................        58.98
Kailua Kona/Waimea Area....................................        89.07
Kauai County...............................................        89.51
Maui County................................................        97.73
Guam.......................................................        82.57
Washington, DC, Area.......................................    100.00\*\
------------------------------------------------------------------------
\*\ By definition, the index of the base area is always 100.00.

    Appendix 6 shows the regression equation in SAS code and the 
regression results. (SAS is a proprietary statistical analysis computer 
software package.)

4.3 Averaging Prices by Item and Area

    After OPM collected, reviewed, and made special adjustments in the 
data (as required), OPM averaged the prices for each item by COLA 
survey area. For example, OPM priced a bag of sugar at three different 
grocery stores in Honolulu County and averaged these prices to compute 
a single average price for sugar in Honolulu. If OPM collected more 
than one price for a particular matched item within the same outlet 
(e.g., priced equivalent brands), OPM used the lowest price by item and 
outlet to compute the average. (The concept is that, if the item and 
brands are equivalent, consumers will choose the one with the lowest 
price.) OPM repeated this item-by-item averaging process for each area.
    For Washington, DC, area prices, we first averaged prices within 
each of the three DC survey areas described in Section 2.5. Then we 
computed a weighted average of the three DC survey areas using census 
data on where Federal employees live as the weights.

4.4 Computing Price Indexes

    Next, OPM computed a price index for each of the items found in 
both the COLA survey area and in the Washington, DC, area. To do this, 
OPM divided the COLA survey area average price by the DC area average 
price and, following the convention used to express indexes, multiplied 
this by 100. For the vast majority of survey items, OPM next applied 
consumer expenditure weights. For a few items, however, OPM first 
applied special processes as described in Sections 4.4.1 and 4.4.2 
below.
4.4.1 Geometric Means
    As described in Section 2.3, OPM selected survey items to represent 
selected detailed expenditure categories (DECs). Generally, OPM 
surveyed only one item per DEC, but in a few cases, OPM surveyed 
multiple items at a single

[[Page 74867]]

DEC. In these cases, OPM computed the geometric mean of the price 
indexes to derive a single price index for the DEC. (A geometric mean 
is the nth root of the product of n different numbers and is often used 
in price index computations.) For example, OPM surveyed two 
prescription drugs--Methylphenidate and Nexium. These two different 
prescription drugs represent a single DEC called ``prescription 
drugs.'' To derive a single price index for the DEC, OPM computed the 
geometric mean of the price index for Methylphenidate and the price 
index for Nexium.
4.4.2 Special Private Education Computations
    As noted in Section 4.2.1, OPM surveyed K-12 private education in 
the COLA and DC areas and computed an average tuition ``price'' that 
reflected all grade levels. Because not everyone sends children to 
private school, OPM made an additional special adjustment for K-12 
education by applying ``use factors.'' These use factors reflect the 
relative extent to which Federal employees make use of private 
education in the COLA and DC areas. For example, Table 5 shows a use 
factor of 2.0302 for Honolulu County. OPM computed this by dividing 
26.86 percent (the percentage of Federal employees in Honolulu County 
with at least 1 child in a private school) by 13.23 percent (the 
percentage of DC area Federal employees with at least 1 child in a 
private school). OPM obtained the percentages from the results of the 
1992/93 Federal Employee Housing and Living Patterns Survey, which is 
the most current comprehensive data available. Table 5 shows the use 
factors and the adjusted price indexes for each COLA survey area.

                          Table 5--Summary of Private Education Use Factors and Indexes
----------------------------------------------------------------------------------------------------------------
                                                  Employees w/children in
                                                      private schools                                Price index
                COLA survey area                --------------------------  Use factor  Price index     w/use
                                                  Local area    DC area                                 factor
----------------------------------------------------------------------------------------------------------------
Honolulu County................................        26.86        13.23       2.0302        78.55       159.48
Hilo Area \*\..................................        18.94        13.23       1.4316        55.57        79.56
Kailua Kona/Waimea \*\.........................        18.94        13.23       1.4316        74.77       107.04
Kauai County...................................        22.46        13.23       1.6977        57.74        98.03
Maui County....................................        20.39        13.23       1.5412        52.76        81.32
Guam...........................................        42.26        13.23       3.1943        52.02       166.18
----------------------------------------------------------------------------------------------------------------
\*\ Use factor data available only for Hawaii County.

4.5 Applying Consumer Expenditure Weights

    Next, OPM applied consumer expenditure weights to aggregate price 
indexes by expenditure group. As noted in Section 2.3, OPM used the 
results of the BLS Consumer Expenditure Survey to estimate the amounts 
middle income level consumers in the DC area spend on various items. 
Using expenditure weights, OPM combined the price indexes according to 
their relative importance. For example, shelter is the most important 
expenditure in terms of the COLA survey and represents about 30 percent 
of total consumer expenditures. On the other hand, the purchase of 
newspapers at newsstands represents less than \1/10\th of 1 percent of 
total expenditures.
    Beginning at the lowest level of expenditure aggregation (e.g., 
sub-PEG), OPM computed the relative importance of each survey item 
within the level of aggregation, multiplied the price index times its 
expenditure percentage, and summed the cross products for all of the 
items within the level of aggregation to compute a weighted price index 
for that level. OPM repeated this process at each higher level of 
aggregation (e.g., PEG and MEG). Appendix 7 shows these calculations 
for each COLA survey area at the PEG and MEG level.
    The above process resulted in an overall price index for each of 
the Pacific COLA areas (shown in Appendix 7), but not for Hawaii 
County, which has two separate COLA survey areas. To compute an overall 
price index for Hawaii County, OPM computed weights based on the number 
of General Schedule (GS) and equivalent Federal employees stationed on 
the Hilo side of the island compared with the number stationed on the 
Kailua Kona/Waimea side of the island. OPM then multiplied each of the 
MEG indexes for Hilo and Kailua Kona by their respective GS employment 
weights and summed the cross products to produce an overall price index 
for Hawaii County. (See Appendix 7.) Table 6 shows the weights OPM 
used.

         Table 6--Hilo and Kailua Kona/Waimea Employment Weights
------------------------------------------------------------------------
                  Area                      GS employment       Weight
------------------------------------------------------------------------
Hilo Area...............................                643         66.7
Kailua Kona/Waimea Area.................                321         33.3
                                         -------------------------------
    Total...............................                964        100.0
------------------------------------------------------------------------

5. Final Results

    To compute the overall living-cost index, OPM added to the price 
index a non-price adjustment factor. The parties in Caraballo 
negotiated these factors to reflect differences in living costs that 
might not be captured by the surveys, and OPM adopted these factors in 
regulation as part of the new methodology. The factor for Honolulu 
County is five index points. The factor for all other COLA areas in 
Hawaii is seven index points. The factor for Guam/CNMI is nine index 
points. The resulting living-cost indexes are shown in Table 7.

              Table 7--Final Living-Cost Comparison Indexes
------------------------------------------------------------------------
                       Allowance area                           Index
------------------------------------------------------------------------
Honolulu County, HI........................................       121.37
Hawaii County, HI..........................................       111.71
Kauai County, HI...........................................       118.14
Maui County, HI............................................       123.62

[[Page 74868]]

 
Guam/CNMI..................................................       119.98
------------------------------------------------------------------------

6. Post Survey Review

    In December 2007, OPM held teleconferences with the COLA Advisory 
Committees in Honolulu, Hilo, Kailua Kona, Kauai, Maui, and Guam to 
review the survey results. We provided the committee members with 
various reports showing the data we collected, examples of how we 
reviewed these data, the data we used in our analyses, and the results 
at the PEG and MEG level, as shown in Appendix 7. We explained how we 
analyzed the rental data and used expenditure weights to combine price 
indexes to reflect overall living costs.

Appendix 1--Prior Survey Results: 1990-2006

------------------------------------------------------------------------
              Citation                             Contents
------------------------------------------------------------------------
73 FR 774..........................  Report on 2006 living-cost surveys
                                      conducted in Alaska.
71 FR 63179........................  Report on 2005 living-cost surveys
                                      conducted in Puerto Rico and the
                                      U.S. Virgin Islands.
70 FR 44989........................  Report on 2004 living-cost surveys
                                      conducted in Hawaii and Guam.
69 FR 12002........................  Report on 2003 living-cost surveys
                                      conducted in Alaska.
69 FR 6020.........................  Report on 2002 living-cost surveys
                                      conducted in Puerto Rico and the
                                      U.S. Virgin Islands.
65 FR 44103........................  Report on 1998 living-cost surveys
                                      conducted in Alaska, Hawaii, Guam,
                                      Puerto Rico, and the U.S. Virgin
                                      Islands.
63 FR 56432........................  Report on 1997 living-cost surveys
                                      conducted in Alaska, Hawaii, Guam,
                                      Puerto Rico, and the U.S. Virgin
                                      Islands.
62 FR 14190........................  Report on 1996 living-cost surveys
                                      conducted in Alaska, Hawaii, Guam,
                                      Puerto Rico, and the U.S. Virgin
                                      Islands.
61 FR 4070.........................  Report on winter 1995 living-cost
                                      surveys conducted in Alaska.
60 FR 61332........................  Report on summer 1994 living-cost
                                      surveys conducted in Hawaii, Guam,
                                      Puerto Rico, and the U.S. Virgin
                                      Islands.
59 FR 45066........................  Report on winter 1994 living-cost
                                      surveys conducted in Alaska.
58 FR 45558........................  Report on summer 1992 and winter
                                      1993 living-cost surveys conducted
                                      in Alaska, Hawaii, Guam, Puerto
                                      Rico, and the U.S. Virgin Islands.
58 FR 27316........................  Report on summer 1993 living-cost
                                      surveys conducted in Hawaii, Guam,
                                      Puerto Rico, and the U.S. Virgin
                                      Islands.
57 FR 58556........................  Report on summer 1991 and winter
                                      1992 living-cost surveys conducted
                                      in Alaska, Hawaii, Guam, Puerto
                                      Rico, and the U.S. Virgin Islands.
56 FR 7902.........................  Report on summer 1990 living-cost
                                      surveys conducted in Alaska,
                                      Hawaii, Guam, Puerto Rico, and the
                                      U.S. Virgin Islands.
------------------------------------------------------------------------

Appendix 2--Estimated DC Area Middle Income Annual Consumer 
Expenditures

              [Asterisks show Detailed Expenditure Categories (DECs) for which OPM surveyed items.]
----------------------------------------------------------------------------------------------------------------
              Level                       Code                Group           Category name       Expenditures
----------------------------------------------------------------------------------------------------------------
1...............................  XTOTAL.............  ..................  Total Expenditure.         $57,910.67
2...............................  FOODTOTL...........  MEG...............  Food..............           6,516.50
3...............................  CERBAKRY...........  PEG...............  Cereals and bakery             426.43
                                                                            products.
4...............................  CEREAL.............  ..................  Cereals and cereal             152.02
                                                                            products.
5...............................  010110.............  ..................  Flour.............               4.76
5...............................  010120.............  ..................  Prepared flour                  12.20
                                                                            mixes.
5...............................  010210.............  ..................  Ready to eat and                95.36
                                                                            cooked cereals *.
5...............................  010310.............  ..................  Rice *............              17.23
5...............................  010320.............  ..................  Pasta, cornmeal                 22.47
                                                                            and other cereal
                                                                            products *.
4...............................  BAKERY.............  ..................  Bakery products...             274.41
5...............................  BREAD..............  ..................  Bread.............              81.05
6...............................  020110.............  ..................  White bread *.....              31.35
6...............................  020210.............  ..................  Bread, other than               49.70
                                                                            white *.
5...............................  CRAKCOOK...........  ..................  Crackers and                    72.78
                                                                            cookies.
6...............................  020510.............  ..................  Cookies *.........              44.31
6...............................  020610.............  ..................  Crackers..........              28.47
5...............................  020810.............  ..................  Frozen and                      20.07
                                                                            refrigerated
                                                                            bakery products *.
5...............................  OTHBAKRY...........  ..................  Other bakery                   100.51
                                                                            products.
6...............................  020310.............  ..................  Biscuits and rolls              37.28
                                                                            *.
6...............................  020410.............  ..................  Cakes and cupcakes              29.32
                                                                            *.
6...............................  020620.............  ..................  Bread and cracker                3.62
                                                                            products.
6...............................  020710.............  ..................  Sweetrolls, coffee              18.16
                                                                            cakes, doughnuts.
6...............................  020820.............  ..................  Pies, tarts,                    12.13
                                                                            turnovers.
3...............................  ANIMAL.............  PEG...............  Meats, poultry,                797.61
                                                                            fish, and eggs.
4...............................  BEEF...............  ..................  Beef..............             216.02
5...............................  030110.............  ..................  Ground beef *.....              90.12
5...............................  ROAST..............  ..................  Roast.............              30.38
6...............................  030210.............  ..................  Chuck roast *.....               8.09
6...............................  030310.............  ..................  Round roast *.....               6.69
6...............................  030410.............  ..................  Other roast.......              15.60
5...............................  STEAK..............  ..................  Steak.............              77.60
6...............................  030510.............  ..................  Round steak *.....              13.00
6...............................  030610.............  ..................  Sirloin steak *...              22.62
6...............................  030710.............  ..................  Other steak.......              41.99
5...............................  030810.............  ..................  Other beef........              17.92
4...............................  PORK...............  ..................  Pork..............             123.62

[[Page 74869]]

 
5...............................  040110.............  ..................  Bacon *...........              24.11
5...............................  040210.............  ..................  Pork chops *......              27.34
5...............................  HAM................  ..................  Ham...............              23.57
6...............................  040310.............  ..................  Ham, not canned *.              22.72
6...............................  040610.............  ..................  Canned ham *......               0.85
5...............................  040510.............  ..................  Sausage...........              22.63
5...............................  040410.............  ..................  Other pork........              25.99
4...............................  OTHRMEAT...........  ..................  Other meats.......              98.39
5...............................  050110.............  ..................  Frankfurters *....              19.93
5...............................  LNCHMEAT...........  ..................  Lunch meats (cold               70.77
                                                                            cuts).
6...............................  050210.............  ..................  Bologna,                        17.74
                                                                            liverwurst,
                                                                            salami *.
6...............................  050310.............  ..................  Other lunchmeats..              53.03
5...............................  LAMBOTHR...........  ..................  Lamb, organ meats                7.69
                                                                            and others.
6...............................  050410.............  ..................  Lamb and organ                   5.87
                                                                            meats.
6...............................  050900.............  ..................  Mutton, goat and                 1.82
                                                                            game.
4...............................  POULTRY............  ..................  Poultry...........             150.59
5...............................  CHICKEN............  ..................  Fresh and frozen               124.71
                                                                            chickens.
6...............................  060110.............  ..................  Fresh and frozen                32.03
                                                                            whole chicken *.
6...............................  060210.............  ..................  Fresh and frozen                92.68
                                                                            chicken parts *.
5...............................  060310.............  ..................  Other poultry.....              25.89
4...............................  FISHSEA............  ..................  Fish and seafood..             174.06
5...............................  070110.............  ..................  Canned fish and                 24.51
                                                                            seafood *.
5...............................  070230.............  ..................  Fresh fish and                  88.71
                                                                            shellfish *.
5...............................  070240.............  ..................  Frozen fish and                 60.84
                                                                            shellfish *.
4...............................  080110.............  ..................  Eggs *............              34.93
3...............................  DAIRY..............  PEG...............  Dairy products....             356.84
4...............................  MILKCRM............  ..................  Fresh milk and                 136.59
                                                                            cream.
5...............................  090110.............  ..................  Fresh milk, all                122.82
                                                                            types *.
5...............................  090210.............  ..................  Cream.............              13.77
4...............................  OTHDAIRY...........  ..................  Other dairy                    220.25
                                                                            products.
5...............................  100110.............  ..................  Butter............              18.06
5...............................  100210.............  ..................  Cheese *..........             111.0