[Federal Register: July 17, 2000 (Volume 65, Number 137)] [Notices] [Page 44103-44173] [[Page 44103]] ----------------------------------------------------------------------- OFFICE OF PERSONNEL MANAGEMENT Report On 1998 Surveys Used to Determine Cost-of-Living Allowances in Nonforeign Areas AGENCY: Office of Personnel Management. ACTION: Notice. ----------------------------------------------------------------------- SUMMARY: This notice publishes the ``Report on 1998 Surveys Used to Determine Cost-of-Living Allowances in Nonforeign Areas.'' The Federal Government uses the results of these surveys 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 Commonwealth of the Northern Mariana Islands, Puerto Rico, and the U.S. Virgin Islands. Based on the survey findings, the Office of Personnel Management is increasing the local retail COLA rate for the Guam and Commonwealth of the Northern Mariana Islands allowance area in an interim regulation published with this notice. This increase is a result of cost-of-living surveys conducted in October and November 1998 using our current methodology for calculating COLA rates. A settlement agreement that is currently awaiting court approval calls for OPM to use a new methodology in conducting future surveys and in calculating future COLA rates. Therefore, the survey results reflected in this rule are not an indication of what survey results or COLA rates would be under the new methodology. DATES: We must receive comments on or before November 14, 2000. ADDRESSES: Send or deliver comments to Donald J. Winstead, Assistant Director for Compensation Administration, Workforce Compensation and Performance Service, Office of Personnel Management, Room 7H31, 1900 E Street NW., Washington, DC 20415-8200; fax: (202) 606-4264; or email: COLA@opm.gov. FOR FURTHER INFORMATION CONTACT: Donald L. Paquin, (202) 606-2838; fax: (202) 606-4264; or email: COLA@opm.gov. SUPPLEMENTARY INFORMATION: Section 591.206(c) of title 5, Code of Federal Regulations, requires the Office of Personnel Management (OPM) to publish nonforeign area cost-of-living allowance (COLA) survey summaries and calculations in the Federal Register. We are publishing the complete ``Report on 1998 Surveys Used to Determine Cost-of-Living Allowances in Nonforeign Areas'' with this notice. In the report, we explain the methodologies, calculations, and findings of the 1998 COLA surveys. Results of Surveys Using an index scale with the living costs in the Washington, DC, area equal to 100, we computed index values of relative living costs in each of the allowance areas. (See the final cost comparison indexes in the Executive Summary of the report.) The results of the surveys show that the local retail COLA rate for the Guam and Commonwealth of the Northern Mariana Islands allowance area should increase from 22.5 percent to 25 percent, the COLA rates for two other areas are currently at the appropriate level, and the COLA rates in eight areas are above the levels indicated by the living-cost indexes. However, the Treasury, Postal Service, and General Government Appropriations Act, 1992 (Pub. L. 102-141), as amended, prohibits reductions in COLA rates through December 31, 2000. Therefore, the interim regulation contains no COLA rate reductions. Corrections to the 1997 Report In preparing our report on the 1998 surveys, we discovered three errors in the 1997 survey report. We discuss these errors below, and we have corrected them in the 1998 analyses and report. We note that these corrections did not affect the COLA rates for any allowance area. Median values for housing. We further analyzed our data on 1997 median housing values and discovered an error that resulted in our publishing incorrect values for Anchorage and Fairbanks in appendix 9. We made these corrections. Although these changes had no effect on the Anchorage index, the Fairbanks index decreased from 107.57 to 107.53. The original and corrected home sales values are as follows: ------------------------------------------------------------------------ Original Corrected ------------------------------------------------------------------------ Anchorage: Lower....................................... $86,859 $86,733 Middle...................................... 119,561 119,236 Upper....................................... 149,073 149,124 Fairbanks: Lower....................................... 78,804 76,086 Middle...................................... 97,110 No change Upper....................................... 122,196 112,128 ------------------------------------------------------------------------ Guam indexes. We had inadvertently omitted cellular phone service when calculating telephone expenses for the homeowner and renter indexes in Guam. We had also used an incorrect value for the Guam personal insurance and pensions index. While correcting these increased the Guam local retail total comparative cost index to 122.63 and the Guam commissary and exchange index to 119.09, these increases did not increase the actual COLA rates for Guam. Renter characteristics and survey communities. In Table 4-2, Housing Profiles, we should have shown one bath instead of two for middle income renters. For upper income renters, we should have shown the rooms as 2-3 bedrooms and 4-5 rooms total. Also, in appendix 11, we should have noted in the footnotes that we relaxed the community specifications for the broker data. We have made these corrections in the 1998 survey report. None of these changes affected the indexes. Comments on 1997 Survey Report OPM published the report on the 1997 surveys for comment in the Federal Register on October 21, 1998 (63 FR 56432). We received five written comments and additional oral comments. Generally, the commenters believed the surveys did not fully consider all expenses incurred in the allowance areas. Some commenters felt the surveys did not account for dissimilarities between the allowance areas and the Washington, DC, area, and that this affected the accuracy of the survey results. OPM recently participated in a major initiative under a memorandum of understanding with plaintiffs in certain COLA litigation. That initiative studied many of these issues. We also engaged in a 2-year partnership pilot project that looked into many of the same issues. We describe these two efforts below, then discuss the substantive comments we received in response to the 1997 survey report. Memorandum of Understanding and Report to Congress In 1996, OPM entered into a memorandum of understanding (MOU) with litigants in the cases of Alaniz v. Office of Personnel Management and Karamatsu v. United States. The MOU committed OPM and the litigants to a ``Safe Harbor'' process for conducting studies relating to the COLA program and the compensation of Federal employees in the allowance areas. The purpose of the Safe Harbor process was to resolve long- contested COLA issues and to assist OPM in preparing a report to Congress on the COLA program. This report, required by the Treasury, Postal Service, and General Government Appropriations Act, 1992 (Pub. L. 102-141), as amended, was due by March 1, 2000. However, the Government and [[Page 44104]] plaintiffs are currently negotiating to settle the contested issues. If the parties achieve settlement, OPM will make many substantive changes in the COLA methodology. Therefore, we have notified Congress that we will report after we conclude the settlement process. During the Safe Harbor process, we avoided making substantive policy changes in the COLA program. We made administrative changes as necessary and implemented other improvements in response to the comments we received. We list these changes in the survey report. COLA Partnership In November 1996, we established a 2-year pilot project to involve agency and employee representatives in a partnership with OPM to help us administer the nonforeign area COLA program. Our goal was to introduce a cooperative effort to help us plan and conduct COLA surveys, explore ways to improve the COLA program, and help everyone, including OPM, better understand issues related to the compensation of Federal employees in the COLA areas. OPM worked with committees established under the pilot project to plan and conduct both the 1997 and 1998 living-cost surveys in the COLA areas. Although the pilot project expired in November 1998, OPM continued to work informally with interested committee members in the analysis of the 1998 survey results. Goods and Services One commenter suggested that we survey costs for building materials such as plywood, framing lumber, cabinets, carpet, and roofing materials. The commenter noted that delivery of these materials to Juneau takes a minimum of 2 weeks, resulting in project delays and higher costs. We currently survey various building material items, including paint, electrical outlets, area rugs, and caulking. We also survey the cost for interior painting and an electrical project, which should reflect higher costs due to material supply delays. Based on this suggestion, we collected prices for plywood on a test basis during the 1998 surveys. We obtained usable data and included these prices in our analysis. The same commenter suggested that OPM consider surveying the cost of an oil change, appliance repair, and dry cleaning. In this and previous surveys, we surveyed both the cost of oil changes and dry cleaning. This year, we added appliance repair as a test item and found that we were able to collect comparable data across areas. Therefore, we used the results of this test item. The commenter also suggested that OPM survey landfill charges for trash and recyclable material disposal. Consumer trash removal is often a tax-supported service or is included in the water-sewer bill. We believe the extent to which consumers pay landfill fees in lieu of higher taxes or utility fees probably differs significantly by area, and we have no information that would allow us to take these differences into consideration. Therefore, we are not adopting this suggestion. One commenter suggested that OPM consider pricing both basic cable TV service and the next higher level of service, at least on a test basis. We adopted the change as a test item for the 1998 survey, but found we could not obtain comparable data across areas. Therefore, we did not use prices for level 1 cable TV service in any area. The same commenter noted that some hospitals in Hawaii have only private rooms, not semi-private rooms as OPM surveyed in 1997. The commenter suggested pricing both private and semi-private hospital rooms. We adopted this change for the 1998 survey. A commenter suggested surveying the price of specially formulated paints that inhibit mildew or pricing mildew additive. For the 1998 survey, we priced mildew additive in each area and added it to the price of a gallon of paint. A commenter suggested OPM add personal computers to the survey. We researched this, but found that it was not feasible to survey comparable brands and models across areas. However, we plan to reconsider surveying this item, perhaps on a test basis, in future surveys. One commenter noted that sales taxes were increasing in Juneau to cover various new facilities and services. We include the local sales tax in the price of items we survey; therefore, the data we use in our price comparisons reflect sales tax increases. The same commenter remarked that the closing of a department store and a pharmacy in Juneau reduced the availability of certain items. The extent to which fewer goods or services leads to higher costs is reflected in the item prices we collect. The availability of goods and services in the allowance areas was one of the research topics under the MOU. One commenter remarked on the frequency of sales in the Washington, DC, area compared to Juneau. In the 1998 and previous surveys, we compared only non-sale prices of identical items from similar outlets. In future surveys, however, we plan to survey the price of the item at the time of the survey. If we adopt this change, we will collect both sale prices and regular prices, depending on whether the item is on sale at the time we visit the outlet. Housing One commenter felt that the median price used by OPM for upper income house sales in Anchorage was too low to be an accurate reflection of prices for upper income homes. The commenter believed that the lower priced homes could not have been in liveable condition or in a safe neighborhood. We used data provided by an Anchorage real estate broker on homes that were sold during the period August 1, 1996, through July 31, 1997. We looked at over 750 upper income home sale prices in south Anchorage, and we believe these produced a representational median. The same commenter recommended that we examine earthquake and flood insurance needs by individual allowance area. In 1992, OPM's contractor for the cost-of-living surveys, Runzheimer International, investigated homeowner/renter insurance coverage for floods and earthquakes in each individual allowance area. Runzheimer found that less than 10 percent of the population in each of the allowance areas purchased these coverages. Because most homeowners and renters do not purchase an earthquake rider, we do not include it in our surveys. Furthermore, whether lenders require homeowners to buy flood insurance depends on where the property is located, and this can be an insurance requirement in any area, including for properties along the rivers and streams in the Washington, DC, area. We are not aware of any data source that would allow us to determine for each survey area the percent of properties in a flood zone.Therefore, we do not survey the cost of this type of coverage. However, we do survey the cost of hurricane and typhoon insurance in tropical COLA areas, where lenders typically require this coverage. Another commenter noted that housing costs are high in Juneau. Our survey of home sales data and other housing expenses in Juneau should capture these costs. A commenter from the Virgin Islands noted that many employees live on the island of St. John. Recognizing that it was not feasible to price all survey items on St. John, the commenter suggested that OPM survey home sales and rental prices and combine these data with the St. Thomas data. We adopted this change for the 1998 survey. [[Page 44105]] Transportation Component Two commenters suggested that OPM reconsider the models of automobiles it prices in the COLA surveys. One commenter suggested that OPM survey more sports utility vehicles. The other suggested that OPM survey a luxury brand, such as BMW. We did not adopt either of these suggestions. We survey three models--Honda Civic, Ford Taurus, and Chevrolet Blazer. These are popular brands and models, although their popularity differs from one area to the next. It was not feasible for us to vary the brands and models by area with the 1998 survey. However, it may be possible to do this in future surveys. As with all survey items, we will consider changing models and brands in future surveys in response to changes in consumer preferences. One commenter believed we should include the cost of windshield repairs in our survey of vehicle repair costs for Alaskans. In the 1997 survey, we surveyed the frequency and cost of windshield replacement in all of the COLA areas and in the DC area. We found that frequency of windshield replacement was greater in Alaska than in the DC area, but that the frequency of windshield replacement in the other COLA areas was about the same as in the DC area. We also found that the cost of windshield replacement in Alaska was greater than the automobile insurance deductible priced in the COLA surveys. Since consumers pay only the deductible for these repairs, we do not need to survey this item. Instead, we add the cost of the deductible to the annual private transportation costs for the Alaska areas. This was done for both the 1997 and 1998 surveys. One commenter suggested that we use the NADA or Kelly Blue Book for the Pacific region to determine the used car values we use in the COLA model. We use the residual value of a car after 4 years to calculate the annual depreciation expense associated with owning an automobile. We currently use books covering the Eastern region. We researched this issue and found that prices in the Pacific region books tend to be slightly higher than in the Eastern region books. However, for administrative simplicity, we did not adopt the proposal because using different residual values for some areas and not others would have significantly complicated the COLA model. The effect of retaining the current practice may slightly overstate living costs in the COLA areas. One commenter noted that airline competition decreased in Juneau. Our survey of airfare costs should capture any higher ticket prices that result from reduced competition. Another commenter suggested that OPM price the cost of an airline ticket purchased 2 weeks in advance. As used in the COLA model, airfares reflect the cost of vacation travel. We researched the availability and prices of airline tickets and found that generally the best deals were available if the ticket was purchased at least 3 weeks in advance and the traveler flew mid-week (i.e., Tuesday through Thursday). Therefore, for the 1998 surveys in both the COLA areas and the Washington, DC, area, we priced the lowest airfares available 3 weeks in advance, departing on a Tuesday and returning on a Thursday, because this best reflects likely vacation travel. Miscellaneous Component Medical care. One commenter felt that medical care in Juneau was limited, resulting in higher health care costs and inferior health care. The commenter said there was a need for costly travel outside the area to obtain some medical services. We currently price a range of medical services within each area, and the COLA model captures any higher local prices. Travel outside the area for medical service is another issue. Some travel may result from an employee's perceptions about the quality of local medical services. We know of no source that allows us to compare objectively the quality of medical services across areas. Therefore, we do not take into account the cost of unreimbursed travel for medical services or any differences in the quality of health care. A commenter from Puerto Rico believed that a major health benefits plan in that area provided a lower level of coverage than most plans in the DC area. The commenter also said the service covered was inconvenient because it required the employee to use preferred providers who often did not accept appointments. Employees had to show up and wait to be seen. The commenter suggested that OPM review and compare the various Federal health benefits plans. We were unable to do this because it would require us to make subjective decisions about what employees do. For example, if an employee chooses a plan that is less convenient or provides a lower level of coverage, the employee accepts inconvenience and lower coverage as a trade-off for the lower insurance premium, presumably with the expectation that the service/ coverage may not be necessary. It is a highly subjective decision that each employee makes. We know of no objective way to quantify this. Another commenter suggested that OPM price psychiatric counseling. We believe it might be feasible to collect prices for this service in each area, but under the current methodology, the weight we would assign it would be very small. (We discuss how we derive and assign weights in section 2.3 of the report.) Therefore, we did not add this item to the survey because it would have increased the administrative and public burden of the survey with little chance of affecting the results. Other Comments Locality pay. One commenter noted that Federal employees in Juneau do not receive the locality pay increases received by employees in the Washington, DC, area. The locality pay law (5 U.S.C. 5304) prohibits the Government from providing locality pay to employees outside the 48 States and the District of Columbia. Retirement. The same commenter was concerned that COLAs do not count for retirement purposes for employees in the allowance areas. Federal law excludes allowances (including COLAs) from basic pay in the computation of retirement annuities. (See 5 U.S.C. 8331(3) and 8401(4).) Office of Personnel Management. Janice R. Lachance, Director. Report on 1998 Surveys Used to Determine Cost-of-Living Allowances in Nonforeign Areas Table of Contents Executive Summary 1. Introduction 1.1 Report Objectives 1.2 The COLA Partnership Pilot Project 1.3 The Safe Harbor Process 1.4 Changes in the 1998 Survey 1.5 Pricing Period 2. The COLA Model 2.1 Measurement of Living-Cost Differences 2.2 Step 1: Identifying the Target Population 2.2.1 Federal Salaries 2.2.2 Federal Employment Weights 2.3 Step 2: Estimating How People Spend Their Money 2.3.1 Consumer Expenditure Survey 2.3.2 Expenditure Categories and Components 2.4 Step 3: Selecting Items and Outlets 2.4.1 Item Selections--The Market Basket 2.4.2 Geographic Coverage and Outlet Selection 2.4.2.1 Geographic Areas 2.4.2.2 Similarity of Outlets 2.4.2.3 Catalog Pricing 2.5 Step 4: Surveying Prices 2.5.1 Data Collection [[Page 44106]] 2.5.2 Inclusion of Sales and Excise Taxes 2.5.3 Surveying the Washington, DC, Area 2.6 Step 5: Analyzing Data and Computing Indexes 2.6.1 Indexes 2.6.2 Item Weights 2.6.3 Category and Component Weights 2.6.4 Computing the Overall Index 3. Consumption Goods and Services 3.1 Categories and Category Weights 3.2 Goods and Services Survey Results 3.2.1 Exchange and Commissary Expenditure Research 4. Housing 4.1 Component Overview 4.2 Housing Model 4.2.1 Expenditure Research 4.2.2 Housing Profiles 4.2.3 Living Community Selection 4.2.4 Housing-Related Expenses 4.2.4.1 Utilities 4.2.4.2 Real Estate Taxes 4.2.4.3 Owners/Renters Insurance 4.2.4.4 Home Maintenance 4.3 Housing Data Collection Procedures 4.3.1 Homeowner Data Collection 4.3.2 Renter Data Collection 4.4 Housing Analysis 4.4.1 Homeowner Data Analysis 4.4.2 Rental Data Analysis 4.5 Housing Survey Results 5. Transportation 5.1 Component Overview 5.2 Private Transportation Methodology 5.2.1 Vehicle Selection and Pricing 5.2.2 Vehicle Trade Cycle 5.2.3 Fuel Performance and Type 5.2.3.1 Impact of Temperature upon Fuel Performance 5.2.3.2 Impact of Road Surface upon Fuel Performance 5.2.3.3 Impact of Gradient Upon Fuel Performance 5.2.3.4 Overall Impact upon Fuel Performance 5.2.4 Vehicle Maintenance 5.2.5 Tires 5.2.6 License and Registration Fees and Miscellaneous Taxes 5.2.7 Depreciation 5.2.8 Finance Expense 5.2.9 Vehicle Insurance 5.2.10 Overall Annual Costs 5.3 Other Transportation Costs--Air Fares 5.4 Transportation Component Analyses 6. Miscellaneous Expenses 6.1 Component Overview 6.2 Component Weights 6.3 Component Categories 6.3.1 Medical Expense Category 6.3.2 Private Education (K-12) Category 6.3.3 Contributions Category 6.3.4 Personal Insurance and Retirement Category 6.4 Miscellaneous Expense Analyses 7. Final Results 7.1 Total Comparative Cost Indexes List of Appendices Appendix 1: Publication in the Federal Register of Prior Survey Results: 1990-1998 Appendix 2: Federal Employment Weights Appendix 3: Consumer Expenditure Surveys Appendix 4: CES Category and Component Expenditures Appendix 5: Item Descriptions Appendix 6: Principal Pricing Changes Appendix 7: Consumption Goods and Services Analysis Appendix 8: OPM Living Community List Appendix 9: Historical Home Market Values and Interest Rates Appendix 10: Historical Housing Data Appendix 11: Summary of Rental Data Analyses Appendix 12: Housing Cost Analysis Appendix 13: Housing Analysis Appendix 14: Private Transportation Cost Analysis Appendix 15: Auto Insurance Calculation Worksheet Special Limits Adjustments Appendix 16: Air Fares Cost Analysis Appendix 17: Transportation Analysis Appendix 18: Transportation Summary Appendix 19: Miscellaneous Expense Analysis--Total Index Development Appendix 20: Miscellaneous Expense Summary Appendix 21: Component Expenditure Accounts Appendix 22: Total Comparative Cost Indexes Executive Summary The Government pays cost-of-living allowances (COLAs) to Federal employees in nonforeign areas in consideration of living costs higher than in the Washington, DC, area. The Office of Personnel Management (OPM) conducts living-cost surveys in order to set the COLA rates. This report provides the results of the 1998 living-cost surveys and compares living costs in the nonforeign COLA areas to those in the Washington, DC, area. We conducted surveys in Alaska, Hawaii, Guam, Puerto Rico, the U.S. Virgin Islands, and the Washington, DC, area. We then analyzed the survey data and produced this report. For the surveys, we contacted about 4,000 outlets and collected approximately 26,000 prices on about 252 items representing typical consumer purchases. We then combined the data using consumer expenditure information developed by the Bureau of Labor Statistics. The final result is a series of living-cost indexes, shown in Table E-1, that compare living costs in the allowance areas to those in the Washington, DC, area. The index for the DC area (not shown) is 100.00 because it is, by definition, the reference area. Table E-1.--Final Cost Comparison Indexes ------------------------------------------------------------------------ Allowance area Index ------------------------------------------------------------------------ Anchorage, Alaska.......................................... 105.65 Fairbanks, Alaska.......................................... 109.19 Juneau, Alaska............................................. 110.46 The rest of the State of Alaska............................ 131.58 City and County of Honolulu, Hawaii........................ 124.51 Hawaii County, Hawaii...................................... 110.89 Kauai County, Hawaii....................................... 117.19 Maui County, Hawaii........................................ 120.32 Guam/CNMI*, Local Retail................................... 125.23 Guam/CNMI, Commissary/Exchange............................. 121.12 Puerto Rico................................................ 105.93 U.S. Virgin Islands........................................ 116.33 ------------------------------------------------------------------------ *CNMI=Commonwealth of the Northern Mariana Islands 1. Introduction 1.1 Report Objectives This report provides the results of the 1998 surveys. Appendix 1 lists previous survey reports and their publication dates. The analyses show the comparative living-cost differences between the Washington, DC, area and the allowance areas listed below. By law, Washington, DC, is the base or ``reference'' area for the nonforeign area cost-of- living allowance program. 1. Anchorage, Alaska 2. Fairbanks, Alaska 3. Juneau, Alaska 4. The rest of the State of Alaska 5. City and County of Honolulu, Hawaii 6. Hawaii County, Hawaii 7. Kauai County, Hawaii 8. Maui County, Hawaii 9. Guam and the Commonwealth of the Northern Mariana Islands (CNMI) 10. Puerto Rico 11. U.S. Virgin Islands 1.2 The COLA Partnership Pilot Project In November 1996, OPM established the COLA Partnership Pilot Project, a 2-year pilot project designed to assist us in administering the COLA program. (See 61 FR 59173.) The pilot project established COLA Partnership Committees and Subcommittees in Alaska, Hawaii, Guam, Puerto Rico, and the U.S. Virgin Islands. Members of the committees and subcommittees included representatives from local area unions and agencies, as well as representatives from OPM. The Committees and Subcommittees worked with OPM in varying degrees to plan the COLA surveys, observe the data collection, and advise OPM on the COLA program and on compensation issues relating to the COLA areas. We have adopted a number of the changes recommended by the Committees and Subcommittees since the start of the project. However, OPM did not renew the COLA Partnership Pilot Project [[Page 44107]] when it expired because we were involved in discussing the nature of future employee involvement in the COLA program as part of the MOU process. The pilot project ended on November 23, 1998. 1.3 The Safe Harbor Process In 1996, we entered into a memorandum of understanding (MOU) with litigants in the cases of Alaniz v. Office of Personnel Management and Karamatsu v. United States. Under the MOU, we committed to a ``Safe Harbor'' process with the litigants to conduct studies relating to the COLA program and the compensation of Federal employees in the allowance areas. The Safe Harbor process had two primary goals: (1) To resolve long-contested issues in the COLA program and (2) to assist OPM in preparing a report to Congress on the COLA program. This report, required by the Treasury, Postal Service, and General Government Appropriations Act, 1992 (Pub. L. 102-141), as amended, was due by March 1, 2000. However, since the Government is currently negotiating to settle several pending court cases in the COLA areas, we will not report to Congress until after the Government concludes these negotiations. 1.4 Changes in the 1998 Survey During the course of the COLA Partnership Pilot Project and the Safe Harbor process, we generally avoided making substantive changes in the COLA program. As with previous surveys, we did make a few non- substantive changes in the 1998 surveys. The majority of these changes related to items or outlets surveyed. (See Appendix 6.) One of the changes was in the Goods and Services Component that involved obtaining more price quotes for each item. In previous surveys, we attempted to get three price quotes (one for each item at three different suitable outlets) for most items in each survey area. In the 1998 survey, we attempted to obtain up to nine price quotes for many items. This significantly increased the number of price observations we used in this survey. 1.5 Pricing Period We traveled to the COLA areas in October and November 1998 to collect the living-cost data. During the same time frame, we collected data in the Washington, DC, area. We collected the prices of some items--those dependent upon the pricing of other items--later. Because we conducted the surveys in October and November, we were not able to collect prices for some winter items, such as downhill skiing. As in previous surveys, we priced some catalog items. We used only catalogs that sell merchandise in both the allowance areas and the Washington, DC, area. To ensure consistent catalog pricing, we used only current catalogs for all catalog items surveyed. 2. The Cola Model 2.1 Measurement of Living-Cost Differences The COLA model measures living-cost differences between the allowance areas and the Washington, DC, area by-- --Selecting typical items that people purchase in these locations, --Calculating their respective cost differences, and --Combining costs according to their relative importance to each other (as measured by relative percentage of expenditures). This involves the following major steps: Step 1: Identify the segment of the population for the target analysis (i.e., typical Federal white-collar employees). Step 2: Estimate how these people spend their money. Step 3: Select items to represent the types of expenditures people usually make and outlets at which people typically make purchases for each selected item. Step 4: Conduct pricing surveys of the selected items in each area. Step 5: Compute price ratios for the surveyed items and aggregate them according to the relative importance of each item. 2.2 Step 1: Identifying the Target Population The study estimates living-cost differences for typical white- collar Federal employees who have annual base salaries between approximately $13,000 and $94,300, the range of the 1998 General Schedule. Because living costs may vary depending on an employee's income level, we analyze living costs at three income levels. 2.2.1 Federal Salaries To determine the appropriate income levels, we-- 1. Analyzed the 1998 distribution of salaries for General Schedule employees in all of the allowance areas combined; 2. Divided this distribution into three income groups of equal size and identified the minimum, maximum, and median salary in each group; 3. Rounded the median values to the nearest $100 to produce the three representative income levels of $23,300, $35,300, and $52,700; 4. Compared living costs at each of these three income levels to produce three sets of estimated expenditures for each allowance area and for the Washington, DC, area; and 5. Combined these estimated expenditures into a single overall index for each allowance area using the employment weights described below. 2.2.2 Federal Employment Weights We used the minimum and maximum values of each income group and the 1998 distribution of General Schedule employees by salary in each allowance area to derive employment weights. We combined these with similar data from 1995 and 1996 to produce a moving average. (We use moving averages to lessen index changes caused by the introduction of new weights over time.) From these averages, we calculated the percentage of the General Schedule workforce in each income group in each area. These percentages became the weights we used to combine estimated expenditures to compute the final index. Appendix 2 shows the General Schedule employment distributions and how we derived the percentage weights. Appendix 21 shows how we used the weights in the final calculations. 2.3 Step 2: Estimating How People Spend Their Money 2.3.1 Consumer Expenditure Survey We base expenditure patterns used in the calculations on national data from the Consumer Expenditure Survey (CES). We obtained from the Bureau of Labor Statistics (BLS) ``prepublished'' CES results for 1994, 1995, and 1997. BLS has advised us that ``prepublished'' CES data may not be statistically significant. To our knowledge, however, it is the only source of comprehensive consumer expenditure information by income level. Therefore, we use it in the model. We use CES data in two ways: (1) To identify appropriate items for the survey and (2) to derive item, category, and component weights. The item weights are not income-sensitive. We analyze aggregated CES data by income level to derive category and component weights. These weights are income-sensitive. Appendices 3 and 4 show the CES data we used in this study. As with the Federal employment weights, we combined the 3 years of CES data to produce a moving average. 2.3.2 Expenditure Categories and Components BLS groups CES items into small, logical families. For example, BLS [[Page 44108]] groups CES pre-published data for beef into four subcategories: Ground beef, roast, steak, and other. BLS further separates the steak and roast groupings into smaller clusters of items (e.g., sirloin and round steak, chuck and round roast). We separated the CES items into the four main cost components specified in our regulations: Consumption Goods and Services, Transportation, Housing, and Miscellaneous Expenses. To develop weighting patterns for the three income levels, we performed linear regression analyses on the CES data shown in Appendix 3.\1\ These analyses produced estimated expenditures at the three income levels identified in section 2.2.1, above. We converted these expenditures to percentages of total expenditures for the four components to produce the values shown in table 2-1. These were the weights we used to combine the expenditures for each of the components into an overall value for each income level in each allowance area and the Washington, DC, area. --------------------------------------------------------------------------- \1\ The midpoint of the moving average of CES data was 1995. Therefore, for the purpose of these regressions, we adjusted Federal salaries to reflect 1995 pay rates. We used the pay increases for 1996 (2.0%), 1997 (2.3%), and 1998 (2.3%) to deflate the 1998 salaries. This produced adjusted Federal salaries of $21,826, $33,071, and $49,326 for use in the regression equations. TABLE 2-1.--Component Expenses Expressed as a Percentage of Total Expenses -------------------------------------------------------------------------------------------------------------------------------------------------------- Goods and 1998 income level 1995 adjusted services Housing Transportation Misc. Total income level* (percent) (percent) (pecent) (percent) (percent) -------------------------------------------------------------------------------------------------------------------------------------------------------- $23,300........................................... $21,826 38.07 26.42 19.24 16.27 100.00 35,300........................................... 33,071 37.48 25.00 19.12 18.40 100.00 52,700............................................ 49,326 36.96 23.72 19.01 20.68 100.00 -------------------------------------------------------------------------------------------------------------------------------------------------------- Note: Values may not total 100 because of rounding. *Income levels are adjusted as described in footnote 1. We further separated Goods and Services Component items into 10 categories and used linear regression techniques to estimate expenditures on these 10 categories by income level. Section 3.1 shows the weights for these categories. We also used the same technique to compute category weights for the Transportation and Miscellaneous Expense Components and to produce ratios of renters to homeowners at each income level. 2.4 Step 3: Selecting Items and Outlets 2.4.1 Item Selections--The Market Basket As noted above, we grouped CES items into ``clusters'' of expenses to determine which items to survey. We chose these clusters so that no market basket item would have an overwhelmingly large or an insignificantly small item weight. For each of these clusters, we identified a set of items to price. Collectively, we call these items a ``market basket.'' Because it would have been impractical to survey each of the thousands of items consumers might buy, the market basket contains representative items. For example, cheddar cheese represents itself and the many other cheeses and related products that consumers purchase. The market basket that we used had approximately 250 items ranging from table salt to new cars to home purchases. Whenever practical, we included in the item description the exact brand, model, type, and size, so that we could price exactly the same items in all areas if possible. For example, we selected a 10.5-ounce can of Campbell's vegetable soup for the survey because it is typical of canned soups, consumers commonly purchase it, and we find it in all areas. Appendix 5 lists the items we survey and their descriptions. Changes in the item list and descriptions are an important aspect of the COLA survey. These changes are necessary to improve the survey and keep the item descriptions current. For this survey, we changed several of the items and descriptions. Appendix 6 lists the major changes and the reason for each. 2.4.2 Geographic Coverage and Outlet Selection Just as it is important to select commonly-purchased items and survey the same items in all areas, it is important to select outlets frequented by consumers and find equivalent outlets in all areas. This involves deciding which geographic areas to survey and which outlets to survey within these geographic areas. 2.4.2.1 Geographic Areas For some areas, the choice of which area(s) to survey was obvious. On St. Thomas, for example, we survey essentially the whole island because the island is not that large, and Federal employees live throughout the island. For other areas, we had to identify specific communities. To do this, we relied mainly on the results of the 1992 Federal Employee Housing and Living Patterns Survey. Among other things, that survey obtained information on where Federal employees lived. We used this information, in consultation with the COLA Partnership Committees and Subcommittees, to select the living communities for pricing housing costs. Again in consultation with the Committees and Subcommittees, we identified outlets within a normal shopping radius of these housing communities. We generally considered outlets within a living community or within an adjoining living community to be within a normal shopping radius. 2.4.2.2 Similarity of Outlets Whenever possible, we (and the Committees/Subcommittees) selected outlets that were popular with consumers and that were comparable to outlets in other areas. For example, we surveyed grocery items at supermarkets in all areas because most people purchase their groceries at such stores and because supermarkets exist in nearly all areas.\2\ The selection of comparable outlets is particularly important because of the significant price variations that may occur between dissimilar outlets (e.g., comparing supermarket prices with convenience store prices). --------------------------------------------------------------------------- \2\ We surveyed groceries at two kinds of supermarkets (i.e., full-service supermarkets and ``warehouse-type'' supermarkets) in areas where both types of supermarkets were common and within a normal shopping radius of the living communities surveyed. We note, however, that some areas do not have warehouse-type supermarkets. We did not survey mebership stores, such as Costco, in any area. --------------------------------------------------------------------------- Although major supermarkets, department stores, and discount stores represented a sizable portion of the survey, we also selected outlets to represent the diversity of consumer shopping options. For example, we could have used department stores for [[Page 44109]] pricing all clothing items. However, this would not have reflected the range of consumer choices. Therefore, we priced some clothing items in department stores, others in shoe stores, others in discount stores, and still others via mail order. For each item, we selected the same type of outlet (e.g., clothing store, discount store, department store) in each area whenever possible. 2.4.2.3 Catalog Pricing We collected 13 item prices by catalog in the survey to reflect this common purchasing option. Catalog pricing also allowed the comparison of items that we would have had difficulty pricing otherwise. We included in the catalog prices any charges for shipping and handling and all applicable taxes. 2.5 Step 4: Surveying Prices As noted earlier, we obtained approximately 26,000 prices on about 250 items from about 4,000 outlets. The 26,000 price observations represents a significant increase over the 1997 survey. In prior surveys, we attempted to get three price quotes (one for each item at three different suitable outlets) for most items in each survey area. In the 1998 survey, we attempted to obtain up to nine price quotes for many items, although we frequently were not able to achieve this goal. Also, there were certain exceptions. For example, we obtained essentially all of the available home sales and rental data meeting the survey specifications. For other items, such as utilities and real estate tax rates, we obtained only one quote in each area because these items have uniform rates within an area. Because the Washington, DC, area has six survey areas, we attempted to get up to nine price quotes for many items in each survey area. 2.5.1 Data Collection To avoid possible conflicts of interest, OPM central office staff collected the price data in each area. In many of the COLA areas, data collection observers, usually designated by the local COLA Partnership Committee or Subcommittee, accompanied our staff. The observers advised and assisted us in contacting outlets, matching items, and selecting substitutes. The observers also advised us on living costs and related compensation issues in their areas. We found this to be a very informative process. We collected most data onsite in stores, repair shops, etc. However, we priced many items, such as insurance, home maintenance services, and private education expenses, by telephone. We collected some items, such as property tax rates, from websites on the Internet. We also purchased home sales and some rental data from various sources. 2.5.2 Inclusion of Sales and Excise Taxes For all items subject to sales and/or excise taxes, we added the appropriate amount of tax prior to analysis. We gathered applicable information on taxes by contacting appropriate sources of information in the allowance areas and the Washington, DC, area. 2.5.3 Surveying the Washington, DC, Area As noted above, we attempted to get more price quotes in the DC area than in the allowance areas because of the size and diversity of the DC metropolitan area and because DC is the basis for all comparisons. For the purposes of the COLA surveys, we divided the DC area into six survey areas: two in the District of Columbia, two in Maryland, and two in Virginia. We surveyed outlets within a normal shopping radius of the housing communities identified in Appendix 8. We combined survey data from each of the six DC survey areas using equal weights. As in the COLA areas, OPM central office staff collected data onsite and by phone in the DC area. Due to funding limitations, allowance area data collection observers did not travel to the DC area to observe and assist in data collection. 2.6 Step 5: Analyzing Data and Computing Indexes 2.6.1 Indexes We derive nonforeign area COLAs from living-cost indexes. These indexes are mathematical comparisons of living costs in the allowance areas to living costs in the Washington, DC, area. An index is a way to state the difference between two prices (or sets of prices). For example, if a can of green beans costs $1.00 in the allowance area and 80 cents in the DC area, canned green beans are 25 percent more expensive in the allowance area than in DC. We can state that difference as a price index of 125. 2.6.2 Item Weights We computed indexes for hundreds of items. As briefly described in section 2.3, we used weights derived from the CES to combine these indexes. These weights reflected the relative amount consumers normally spend on different items. For example, the price of a can of green beans has a lower weight than the price of a pound of apples because, according to the CES, people generally spend less on canned green beans than on apples. (People typically buy more apples than green beans.) The COLA model uses a fixed-weight indexing methodology. The model bases the weights used on the expenditure patterns of consumers nationwide as reported by the CES. This is the only source we are aware of that provides expenditure information by income level. 2.6.3 Category and Component Weights As described in section 2.3.2, we also computed income sensitive category and component weights. This allowed us to combine comparative price data in a manner that reflected the spending patterns of people at each income level. The way we combined data varied among the components. For the Goods and Services and Miscellaneous Expense Components, we combined indexes within each category using the CES weights to derive an overall index for the category. We then combined the category indexes into an overall component index using the income-sensitive category weights described above. For the Transportation and Housing Components, we used the same approach in combination with a cost-build- up approach. For example, we computed the annual cost of owning and operating an automobile by taking individual prices (e.g., automobile financing, insurance, gas and oil, and maintenance) and computing an overall dollar cost for each area. We compared these costs with those in the DC area to compute the Private Transportation Category index. We then combined this index with the Other Transportation Category index using income sensitive category weights to compute an overall Transportation Component index for each area. 2.6.4 Computing the Overall Index We combined the item, category, and component indexes using the process prescribed in section 591.205(c) of title 5, Code of Federal Regulations. This is a five-step process that involves converting the indexes to dollar values, which we then weight, combine, and compare to compute a final weighted-average index. We describe the process in detail below. First, we used the CES data and the income ranges described in section 2.2.1 to determine how much money consumers typically spend on each component at each income level. These amounts appear in the table below and in Appendix 21. We derived the amounts by taking the component weights shown in Table 2-1 and multiplying them times the [[Page 44110]] representative income levels described in section 2.2.1. Table 2-2.--Typical Consumer Expenditures by Income Level and Component ---------------------------------------------------------------------------------------------------------------- Goods and Income level services Own/rent Transportation Misc. Total ---------------------------------------------------------------------------------------------------------------- Lower........................................... $8,870 $6,156 $4,483 $3,791 $23,300 Middle.......................................... 13,230 8,825 6,749 6,495 35,300 Upper........................................... 19,478 12,500 10,018 10,709 52,700 ---------------------------------------------------------------------------------------------------------------- Note: Values may not total because of rounding here and in Table 2-1. Second, for each allowance area, we multiplied the dollar values above by the component indexes for the allowance area. Because the housing component consisted of two indexes (one for owners and another for renters), we produced total relative costs separately for owners and renters. Third, for each allowance area and income level, we combined the total relative costs for owners and renters using as weights the proportion of owners and renters as identified in the CES. (See section 4.2.1.) This produced an overall expenditure dollar amount for each income level in each allowance area. Fourth, we computed a single overall average expenditure for each allowance area by combining the income level expenditures using the allowance area General Schedule employment distribution as weights. This produced a single overall dollar expenditure value for the allowance area. Using the same General Schedule employment weights, we also computed a single overall dollar expenditure value for the DC area. The final step was to divide the overall dollar expenditure for the allowance area by the overall dollar expenditure for the DC area to compute a final index. The last section of this report and Appendix 22 show these indexes. 3. Consumption Goods and Services 3.1 Categories and Category Weights Based on the CES data, we identified 10 categories of expenses within the Goods and Services Component. Using linear regression analyses and the CES data, we identified the portion of total Goods and Services expenditures that the typical consumer spends in each category at various income levels. Table 3-1 shows the categories and the relative expenditures. Table 3-1.--Category Weights Expressed as a Percentage of Goods and Services Expenditures By Income Level ------------------------------------------------------------------------ Income levels Category -------------------------------------- Lower Middle Upper ------------------------------------------------------------------------ Food at home..................... 27.03 24.05 21.30 Food away from home.............. 13.43 14.18 14.87 Tobacco.......................... 2.82 2.34 1.90 Alcohol.......................... 2.33 2.40 2.47 Furnishings & household 15.36 16.64 17.82 operations...................... Clothing......................... 13.02 13.50 13.94 Domestic service................. 1.73 1.95 2.15 Professional services............ 7.09 6.82 6.57 Personal care.................... 3.91 3.77 3.64 Recreation....................... 13.27 14.35 15.34 &radic -------------------------------------- Totals....................... 100.00 100.00 100.00 ------------------------------------------------------------------------ Note: Values may not total 100 because of rounding. 3.2 Goods and Services Survey Results Section 2.6 of this report provides a detailed explanation of the economic model used to analyze the price data. As it applies to Goods and Services, the approach involved comparing the average prices of market basket items in each allowance area with those in the Washington, DC, area. We aggregated the resulting price ratios into subcategory and then category indexes using the moving-average expenditure weights derived from the CES data. Appendix 7 shows for each allowance area 10 category indexes, the weights used at each of the 3 income levels, and the overall Goods and Services Component indexes. The appendix does not include the Washington, DC, area because it is, by definition, the reference area. Therefore, the DC indexes are 100. 3.2.1 Exchange and Commissary Expenditure Research Executive Order 10000, as amended, requires OPM to adjust COLA rates when employees have special purchasing privileges, such as unlimited access to commissaries and exchanges. In Guam, some employees have such access, so we priced the same market basket of Goods and Services items at the commissaries and exchanges in Guam as we used for the local retail pricing. We obtained one price quote for each market basket item found in these facilities. Employees who have access to military facilities make some of their purchases in these facilities and make other purchases elsewhere. Therefore, we used the results of a survey of Federal employees to determine the percentage of purchases that families typically make in military facilities versus local outlets. For example, as Table 3-2 shows, we estimated that employees with commissary/exchange [[Page 44111]] access in Guam purchase approximately 70 percent of their Food at Home items at a commissary and purchase the remaining 30 percent in local retail outlets. Table 3-2.--Percentages of Purchases Made at the Commissaries and Exchanges in Guam ------------------------------------------------------------------------ Category Percentage ------------------------------------------------------------------------ Food at home............................................... 70.0 Food away.................................................. 0.0 Tobacco.................................................... 64.0 Alcohol.................................................... 76.0 Furnishings & hsld. oper................................... 64.5 Clothing................................................... 43.7 Domestic service........................................... 0.0 Professional services...................................... 0.0 Personal care.............................................. 49.3 Recreation................................................. 49.7 ------------------------------------------------------------------------ We used these percentages to aggregate the local retail and commissary/exchange prices into one set of appropriate, blended prices, which we refer to as the Commissary/PX prices. We compared the blended prices to the local retail prices in the Washington, DC, area to compute Commissary/PX Goods and Services Category indexes. We then combined these indexes using CES weights to derive an overall Commissary/PX Goods and Services Component index. Just as with the Guam Local Retail Goods and Services Component index, we combined the Guam Commissary/PX Goods and Services Component index with the indexes for the Housing, Transportation, and Miscellaneous Expense Components to derive a single, overall Commissary/PX index for the Guam allowance area. 4. Housing 4.1 Component Overview The Housing Component consists of the following expenses related to owning or renting a dwelling: Mortgage or rent payments, Utilities, Real estate taxes, Homeowner's or renter's insurance, Home maintenance, and Telephone expenses. At each of the three income levels, we measured the annual housing costs for homeowners and renters separately. We then combined the results using as weights the percentages of owners and renters reported by the CES. 4.2 Housing Model 4.2.1 Expenditure Research We used the CES to determine the national average ratio of families who own, as opposed to rent, their residences at each income level. Using the tenure data by income range as input into a linear regression analysis, we calculated the owner and rental weights shown in Table 4-1 and in Appendix 22. We excluded data for homeowning families without a mortgage because they were not typical of Federal homeowners in the base area or in the allowance areas. Table 4-1.--Owner/Renter Weights ------------------------------------------------------------------------ Income levels -------------------------------------- Category Lower Middle Upper (percent) (percent) (percent) ------------------------------------------------------------------------ Homeowner with mortgage.......... 37.96 47.26 60.70 Renter........................... 62.04 52.74 39.30 -------------------------------------- Totals....................... 100.00 100.00 100.00 ------------------------------------------------------------------------ We also used the CES data to identify which home-maintenance items to price and to establish the relative importance of those items. 4.2.2 Housing Profiles To compare housing costs in all locations, we used six typical housing profiles--three for homeowners and three for renters. Table 4-2 shows these profiles. We assigned one owner and one renter profile to each income level. We attempted to collect information on the living area, numbers and types of rooms, and other information that might influence home sale or rental prices. This information was rarely available for rental units, so we relied on bedroom count and living community to segregate rental prices by income level. We used the additional information shown in Table 4-2, however, during the interview of rental brokers to collect broker data. Information about characteristics of houses sold was also difficult to collect on a consistent basis across all areas. Although detailed information about the houses sold was available for many areas, it was not available for other areas, including the District of Columbia and the Maryland suburbs of the Washington, DC, area. The only housing characteristics that were consistently available across all areas were house type and size. We surveyed only the prices of single family detached houses in each area and relied mainly on house size and living community to segregate home sales by income level. As shown in Table 4- 2, these size ranges overlap. Therefore, when we priced housing in the same living community at two or more income levels, we used the additional information to separate home sales observations into the appropriate income level so that no single home sale observation appeared at more than one income level. Table 4-2.--Housing Profiles ---------------------------------------------------------------------------------------------------------------- Renters Owners ------------------------------------------------------------------------------- Income level Additional Additional Key characteristic information Key characteristic information ---------------------------------------------------------------------------------------------------------------- Lower........................... 1 bedroom 3 rooms total, 1 Detached house, 4 rooms total, 2 apartment. bath; reference 600 to 1,200 sq. bedrooms, 1 bath; size: 600 sq. ft. ft. reference size: 900 sq. ft. Middle.......................... 2 bedroom 4 rooms total, 1 Detached house, 5 rooms total, 3 apartment. bath; reference 1,000 to 1,600 bedrooms, 1 bath; size: 900 sq. ft. sq. ft. reference size: 1,300 sq. ft. [[Page 44112]] Upper........................... 2 bedroom 4 rooms total, 2 Detached house, 7 rooms total, 3 townhouse or baths; reference 1,400 to 2,300 bedrooms, 2 detached house. size: 1,100 sq. sq. ft. baths; reference ft. size: 1,700 sq. ft. ---------------------------------------------------------------------------------------------------------------- We use the reference sizes in Table 4-2 for the calculation of utility costs in the model. (See section 4.2.4.1.) As noted above, they are not the only sizes surveyed for each profile. 4.2.3 Living Community Selection As discussed briefly in section 2.4.2.1, we identified the living communities for the survey based on the results of the 1992 Federal Employee Housing and Living Patterns Survey and in consultation with the COLA Partnership Committees and Subcommittees. Appendix 8 identifies the survey communities. As with previous surveys, we identified nine homeowner and nine renter communities for the Washington, DC, area--one for each income level in each of the three areas (DC, Maryland, and Virginia). In the allowance areas, we identified up to three homeowner and three renter communities--one for each income level. We could not achieve the three-community owner/renter goal in many of the allowance areas because of the relatively few home sales and rental opportunities or data availability in these areas. In such areas, we collected prices for the entire survey area or allowance area rather than in specific communities. We did this in Fairbanks, Juneau, Nome, Hilo, Kailua Kona, Kauai, Maui, Guam, St. Croix, and St. Thomas/ St. John. In these areas, we included all home sales and/or rental rates meeting the housing characteristics for the particular income group in the analysis.\3\ --------------------------------------------------------------------------- \3\ In Puerto Rico we were able to obtain relatively few broker rental quotes for the communities identified in the survey specifications. Therefore, we relaxed the community specifications and used broker rental data for all communities in the greater San Juan metropolitan area. --------------------------------------------------------------------------- For most areas in which we identified discrete living communities, we used zip code boundaries. The exceptions were Anchorage and San Juan. In Anchorage, we used the multiple listing service location codes that realtors commonly use in that area. In San Juan, we used the name of the municipio or community. 4.2.4 Housing-Related Expenses Based on the CES data, we categorized housing-related expense items into one of five groups in the COLA model. These groups were-- --Utilities, --Real estate taxes, --Owners/renters insurance, --Maintenance, and --Telephone expenses. 4.2.4.1 Utilities Electricity, oil, gas, and water. Many utility companies were able to provide current charges per unit of consumption and average consumption patterns for all households. The companies were not, however, able to provide separate consumption patterns by the size or type of housing. Because many utility costs vary by size of house, we needed a factor to derive the utility rates at each of the home profiles. Table 4-3 shows the standard square foot sizes and utility factors used for each home profile. We calculated the factors by assuming that utility use increases or decreases at half the rate that square footage increases or decreases. Table 4-3.--Utility Factors ---------------------------------------------------------------------------------------------------------------- Renter profile Owner profile Income level --------------------------------------------------- Sq. ft. Factor Sq. ft. Factor ---------------------------------------------------------------------------------------------------------------- Lower....................................................... 600 .73 900 .85 Middle...................................................... 900 .85 1,300 1.00 Upper....................................................... 1,100 .92 1,700 1.15 ---------------------------------------------------------------------------------------------------------------- In each area, we obtained the price of each of the types of utilities noted above. Where available, we also gathered from local utility companies information on average annual consumption data per household. We used the local rates and consumption information to compute average annual utility costs. We then used the above factors to adjust the total annual utility costs for each of the various housing profiles. In the DC area, we were unable to obtain estimates for electricity usage for houses heated by gas or oil. However, we were able to obtain kilowatt usage for all-electric houses. In order to avoid potential double counting of utility costs, we used the all-electric data for the DC area. Double counting utility costs was not a problem in the warm- area COLA areas, where there is little heat expense. It also was not a problem in Alaska, where most consumers use gas or oil heat, not electric heat. In the Alaska surveys, we price gas or oil in addition to electricity. Telephone. Telephone expenses consisted of local service charges, additional charges for local calls (if applicable), charges for long distance calls, and basic cellular phone service. To measure estimated expenses for local service and local calls, we surveyed the cost of touch-tone service with unlimited calling in each area. To estimate long distance charges in all areas, we priced from a major long distance provider the cost of three 10-minute direct dial calls per month to large U.S. mainland cities (Los Angeles, Chicago, and New York). As in previous surveys, we priced a call placed in the survey area at the time of day necessary to be received in the respective city at 8:00 p.m. local time. In many areas, this resulted in pricing a combination of daytime and evening-rate calls. We also priced the basic monthly plan for cellular phone service in each area. We derived weights from CES data to [[Page 44113]] account for the portion consumers spend on regular phone service and cellular phone service. We then used these weights to combine the prices of these two types of phone service. 4.2.4.2 Real Estate Taxes For this study, we contacted the local tax assessors or municipal websites on the Internet to obtain real estate tax information on the living communities surveyed. We applied these real estate tax formulas to the median home values for each income level to estimate annual real estate taxes. 4.2.4.3 Owners/Renters Insurance We gathered homeowners' insurance rates for each of the survey areas for both renter and owner profiles. For renters, we used the following estimated content values: $25,000 at the lower income level, $30,000 at the middle income level, and $35,000 at the upper income level. We raised the values for the middle and upper income levels this year after examining test data collected during the 1997 surveys at the request of the Guam COLA Partnership Committee. For homeowners, the cost of insurance was dependent on the median home values calculated as part of this survey. In most areas, we assumed that the structure was equal to 80 percent of the total home value. In Hawaii, where the land represents a greater proportion of property value, we used 50 percent. We priced hurricane insurance in all of the Hawaii allowance areas, Guam, Puerto Rico, and the U.S. Virgin Islands. In research previously conducted for OPM, the contractor found that homeowners and renters rarely purchased insurance coverage for other disasters, such as floods and earthquakes, in any of the allowance areas. (See section 4.2.4.3 of the Report to OPM on Living Costs in Selected Nonforeign Areas and in the Washington, DC, Area, December 10, 1992, at 57 FR 58556.) Insurers we contacted in the 1998 survey indicated that this is still the case. Therefore, we did not survey additional riders for flood or earthquake insurance. 4.2.4.4 Home Maintenance We computed estimated home maintenance expenses for each of the homeowner and renter profiles. We derived separate home maintenance expenditure amounts for both owners and renters from the CES. Not surprisingly, the CES indicates that renters spend relatively little on home maintenance compared with homeowners. As done in previous surveys, we priced both home maintenance services as well as home maintenance commodities using the CES information to identify items to price and the weights associated with these items. The maintenance service items priced were interior painting, plumbing repair, electrical repair, and pest control. In the Nome area, however, we did not price pest control because local sources indicated it is not necessary. The maintenance commodities priced were bathroom caulking, a kitchen faucet set, an electrical outlet, latex interior paint, and a fire extinguisher. To compute home maintenance cost differences between each allowance area and the Washington, DC, area for the homeowner and renter profiles, we computed an index for each maintenance item by comparing the allowance area price to the DC area price. As with the Goods and Services Component items, we used the CES data to weight these maintenance indexes into an overall home maintenance index for each area. To combine the maintenance indexes with the other homeowner and renter costs, which were expressed in dollar amounts, we converted the indexes to dollars. We did this by multiplying the index for each area by the average maintenance expense reported in the CES for owners and renters. We assigned this cost to the middle-income homeowner and renter profile. Logically, maintenance costs for larger homes would generally be greater than costs for middle-sized homes, while costs for smaller homes would generally be less. Therefore, we applied the same owner and renter multipliers used in the utilities model to recognize differences in maintenance costs due to house size at the various income levels. 4.3 Housing Data Collection Procedures We collected home sales information from multiple listing type services and rental information mainly from rental brokers and advertisements. 4.3.1 Homeowner Data Collection We obtained the selling prices of homes that matched the housing profiles in each living community for home sales that occurred roughly during the 12-month period preceding and including the survey month. The amount of data obtained depended on the number of home sales in the community and the availability of square footage and other information on housing characteristics. This in turn depended on the size of the community, economic conditions, the quality and quantity of realty data available, and the willingness and ability of local realty professionals to provide data. We obtained relatively large quantities of home sales data in all areas except Nome. In Nome, home sales were extremely limited because Nome is not very large. In previous surveys, we also obtained relatively little data in St. Thomas. This year, we obtained and used housing data for both St. Thomas and St. John. Also, with the assistance of the Virgin Islands Assessor's Office, we obtained significantly more data than we have been able to get in previous years. These data identified houses that had been significantly damaged by hurricanes or other factors, and we excluded these from our calculations. Identifying houses that were uninhabitable, severely damaged, or otherwise in need of significant repairs was impossible for most areas, given the limited amount of information available from the listing services. As discussed in section 4.4.1 below, we use the median rather than the average home value to compute housing costs. (The median is the middle value in a rank-ordered set of observations and tends to be less sensitive than the average to unusually low or high values at the ends of a range of data.) Nevertheless, in some of the databases we purchased, the quantity of exceptionally low priced homes had a significant effect on the median. Therefore, in all areas, we trimmed home sale prices that were less than $30,000, recognizing that $30,000 was probably a conservative price threshold for most areas. We trimmed homes of $1,000,000 or more at the upper level. We also trimmed properties of 1 acre or larger. 4.3.2 Renter Data Collection We also obtained rental data from a variety of sources, e.g., brokers, rental management firms, property managers, newspaper advertisements, and other listings. Analyses of these data revealed what appeared to be two separate rental markets: A broker market and a non-broker market. Rental rates and estimates provided by brokers generally exceeded those obtained from other sources. We discuss the methodology used to analyze these two data sets in section 4.4.2. 4.4 Housing Analysis 4.4.1 Homeowner Data Analysis One of the most important factors relating to the price of a home is the number of square feet of living space. For each income profile in each allowance area and the Washington, DC, [[Page 44114]] area, we computed price per square foot for each of the comparables and determined the median price per square foot. We use the median to reduce the volatility of the housing data from one survey to the next because a relatively few extremely high or low home prices could significantly influence average housing prices. We then multiplied the median price per square foot by the reference square footage for the income level to determine the home purchase price. As was done in the last survey, we also used historical housing data in addition to data collected in this survey. Appendix 9 shows these data. For all areas except Oahu, the historical data are from previous living-cost surveys that were published in the Federal Register beginning with the 1990 report. (See Appendix 1 for a listing of these publications). The data for the period prior to 1990 were published with the results of the 1991-1992 living-cost surveys at 57 FR 58617 (December 10, 1992) . All housing values are based on the community selections and analytical methodologies used at the time of each respective survey. For Oahu, we surveyed housing prices in new living communities beginning with the 1997 surveys. Because our historical data did not cover these communities, we obtained additional historical price data for use in our 1997 and subsequent survey analyses. The historical housing data used were estimated annual principal plus interest payments by income level in each area. To combine these data, we used weights that we derived from the 1992 Federal Employee Housing and Living Patterns Survey. These weights reflect the proportion of Federal employee homeowners by year of purchase in all allowance areas and in the Washington, DC, area. Appendix 10 shows the historical housing weights and analyses. 4.4.2 Rental Data Analysis We assigned each rental quote to a single income level based on the criteria shown in Table 4-2. As discussed earlier, we received rental data from both broker and non-broker sources. In each area, the quantity of data obtained from either source varied significantly. Therefore, we found that analyzing all of the rental data (both broker and non-broker) together for an area and income level was undesirable. Instead, we analyzed broker and non-broker data separately by income level. As with the housing data analyses, we used the median rental values. For each income level, we separately ranked rental rates from low to high for broker and non-broker data. We determined the median values for broker and non-broker data for each group and then averaged them to compute a single rental value for each income level. Because we have no information on how the Federal employees who rent generally secure their lodgings, we applied equal weights to the broker and non- broker data to compute an overall average rental rate for the area and income level. Because there was insufficient non-broker data in the unfurnished rental units category, we used partly furnished and unfurnished units in the Hawaii areas. Similarly, we used apartment and furnished units in St. Croix at the middle and upper income levels because no other data were available. Appendix 11 shows the broker and non-broker medians and final results. As noted in the appendix, we found inexplicable rental price trends in some of the data, particularly in the broker data. Therefore, as we explain in the footnotes of the appendix, we adjusted the rental data to address these anomalies. 4.5 Housing Survey Results In the above sections, we described the processes used for determining the costs for maintenance, insurance, utilities, real estate taxes, rents, and homeowner mortgages. Appendix 12 shows the cost of each of these items for renters and homeowners in each allowance area and in the Washington, DC, area. Appendix 13 compares the total cost of these items by income level in each allowance area with the total cost of the same items by income level in the DC area. Again, there are separate comparisons for renters and homeowners. The final housing-cost comparisons take the form of indexes that are used in Appendix 21 to derive the total, overall indexes for owners and renters. 5. Transportation 5.1 Component Overview The transportation component consists of two categories: Automobile Expense and Other Transportation Costs. The Automobile Expense Category reflects costs relating to owning and operating a car in each area. The Other Transportation Costs Category is represented by the cost of air travel from each location to common points within the contiguous 48 States. 5.2 Private Transportation Methodology As in previous surveys, we analyzed automobile transportation costs for three commonly purchased vehicles: A domestic auto, an import auto, and a utility vehicle. We used new car costs for these analyses because we believe pricing used vehicles of equivalent quality in each area would require value judgments that could introduce inconsistencies. 5.2.1 Vehicle Selection and Pricing We surveyed the same three models of automobiles in all areas-- --Domestic: Ford Taurus SE 4-door sedan 3.0L 6 cyl. --Import: Honda Civic DX 4-door sedan 1.5L 4 cyl. --Utility: Chevrolet S10 Blazer 4X4 2 door 4.3L 6 cyl. For each model car, we collected new vehicle prices at dealerships in each area. All vehicles had standard options, such as automatic transmission, AM/FM stereo radio, and air conditioning. In Alaska locations, we included special additional equipment (i.e., engine-block heaters and heavy-duty batteries) in new-vehicle prices. We also priced snow tires in Alaska. (See section 5.2.5.) In addition to the manufacturer's suggested retail price, the price included additional charges such as shipping, dealer preparation, additional dealer markup, excise tax, sales tax, documentation fees, and any other one-time taxes or charges. We encountered problems in obtaining comparable car sales data in each area because of survey timing. As stated in section 1.5, we conducted the survey in October and November 1998, when the dealers were just beginning to receive shipments of the new 1999 models. However, not all dealers had the models we were surveying. Therefore, we obtained the prices of both the 1998 and 1999 models (to the extent the 1999 prices were available). Not surprisingly, we discovered that many dealers were charging significant markups for the 1999 models and significantly reducing or eliminating markups on the 1998 models. We found this in many areas, including the Washington, DC, area. Because we had only 1998 model prices across all areas, we used the 1998 model prices instead of the 1999 model prices. To overcome the problem caused by the usual dealer markups, we used the dealer markup for the same brands surveyed in the 1997 survey on the premise that these markups, which were obtained in the summer of 1997, were more typical. 5.2.2 Vehicle Trade Cycle Calculating the cost of owning and operating a vehicle requires knowing the mileage and period of ownership. The automobile industry uses the term [[Page 44115]] ``trade cycle'' to describe these two factors. The trade cycle is the length of time (in months or years) and the total number of miles driven in that time period. The OPM model uses this information to compute annual costs related to fuel, oil, tires, maintenance, and depreciation. As with the previous living-cost analyses, we used a 4- year, 60,000-mile trade cycle in all areas. 5.2.3 Fuel Performance and Type All vehicles in the 1998 study used regular unleaded fuel. We collected self-service cash prices of unleaded regular gasoline at name-brand gas stations in the Washington, DC, area and in all allowance areas. In Alaska, we surveyed both self-serve and full-serve gas prices. To establish average fuel-performance ratings, the COLA model uses the ``city driving'' figures published by the U.S. Environmental Protection Agency (EPA). The model uses the ``city'' figures instead of ``highway'' figures because all locations contained considerable stop- and-go driving conditions or required cautious driving because of poor road conditions. As in previous COLA surveys, we included in our analysis the following fuel-performance factors: temperature, road surface, and gradient. OPM conducted previous research to determine these factors. We discuss this research and the factors below. 5.2.3.1 Impact of Temperature Upon Fuel Performance Temperature affects gas mileage. The lower the temperature, the fewer miles-per-gallon achieved, and vice versa. According to EPA's Passenger Car Fuel Economy: EPA and Road, the temperature at which no adjustments to fuel performance occur is 77 deg.F. Below that temperature, miles-per-gallon achieved drops. Above 77 deg.F miles-per- gallon achieved improves. The model uses the average monthly temperatures for each allowance area and the DC area as reported in The Weather Almanac, published by Ruffner and Blair. For each location and month, the model uses the appropriate factor from the EPA study based on the average monthly temperature for the area. We then average these factors to derive a single overall factor for each location. Table 5-1 shows the results of these calculations. 5.2.3.2 Impact of Road Surface Upon Fuel Performance The model assumes that Federally controlled roadways are typically composed of concrete and/or high-load asphalt and that locally controlled roadways are typically composed of low-load asphalt. EPA's research indicates that cars are generally more fuel-efficient on the firmer, high-load surfaces than on the softer, low-load surfaces. Although traffic patterns and road usage vary among areas, previous research conducted for OPM produced no relevant findings regarding this issue. Therefore, the model uses the assumption that Federally- controlled roadways generally support twice the traffic of, or are used at least twice as much as, locally controlled roadways. In each allowance area, we collected the total mileage falling into either the Federal or local categories. For example, Alaska contains 5,512 miles of Federally controlled roads and 7,120 miles of locally controlled roads. The usage assumption increased Federal road mileage by a factor of two for the Alaska allowance areas. We applied the average low-load asphalt factor (which reflects dry, wet, and snowy conditions) to the local mileage percentage and the average concrete and/or high-load asphalt factor to the Federal mileage percentage. This produced two weighted average factors--one for the Alaska allowance areas and another for the other allowance areas. Table 5-1 shows these factors. We assigned the Washington, DC, area a factor of 1.00 on the premise that the vast majority of traffic in that area travels on dry, high-load surfaces. Section 5.2.3.4 describes the application of these factors. 5.2.3.3 Impact of Gradient Upon Fuel Performance We also estimated the effect of gradient on gas mileage from EPA's Passenger Car Fuel Economy: EPA and Road. Local topography (i.e., gradient) affects fuel efficiency. EPA provides mileage factors based upon various gradients ranging from less than 0.5 percent (essentially flat) to greater than 6 percent (steep). In research previously conducted for OPM, the contractor reviewed the topographic features of each area and found a wide range of road conditions. However, the contractor was unable to find relevant information on the types of terrain drivers typically encounter in each area or the number of miles drivers travel in each type of terrain. Lacking such information, the contractor assumed that drivers in the allowance areas generally traveled roads having approximately the same gradients that are found on average in the United States. Applying the information from EPA's research, we computed a fuel- performance factor of 0.98 for this type of driving. We assigned this factor to each allowance area. For the DC area, we used a factor of 1.00 on the premise that the vast majority of traffic in that area travels on major freeways and highways that are relatively flat. The next section describes the application of these factors. 5.2.3.4 Overall Impact Upon Fuel Performance We applied the factors described above to make adjustments in the average gas mileage ratings for each type of automobile surveyed for each allowance area and for the Washington, DC, area. The adjustment factors compound; that is, the total adjustment is the result of multiplying the three individual factors together for each area. In Table 5-1, the factor 1.00 means that no adjustment in EPA fuel performance is appropriate. A factor of less than 1.00 means that the estimated gasoline mileage in the area is less than the EPA average. For example, the total adjustment factor for Juneau is 0.84. This means that the estimated gasoline mileage in Juneau is 84 percent of the EPA estimated average. Note that the adjustment factor for the DC area (0.94) indicates that average gasoline mileage in that area is also below the EPA estimate. Table 5-1.--Summary of Fuel-Performance Adjustments ---------------------------------------------------------------------------------------------------------------- Road Location Temperature surface Gradient Total ---------------------------------------------------------------------------------------------------------------- Anchorage................................................... 0.88 0.96 0.98 0.83 Fairbanks................................................... 0.85 0.96 0.98 0.80 Juneau...................................................... 0.89 0.96 0.98 0.84 Nome........................................................ 0.85 0.96 0.98 0.80 Hawaii...................................................... 0.99 0.98 0.98 0.95 Virgin Islands.............................................. 1.01 0.98 0.98 0.97 [[Page 44116]] Puerto Rico................................................. 1.01 0.98 0.98 0.97 Guam........................................................ 0.99 0.98 0.98 0.95 Washington, DC.............................................. 0.94 1.00 1.00 0.94 ---------------------------------------------------------------------------------------------------------------- 5.2.4 Vehicle Maintenance We surveyed the cost of common maintenance services and repairs performed on the vehicles surveyed. The services and repairs were: Tuneup Oil change Automatic transmission fluid change Flush/fill coolant Muffler/exhaust pipe replacement Constant velocity joint (CVJ) boot replacement Windshield replacement We used the automobile manufacturers' recommended mainte-nance schedules to determine the frequency of performing each of the first five maintenance jobs. Maintenance schedules vary, depending on the driving conditions typically encountered. Consistent with the assumptions used for fuel economy and tire mileage, we assumed that driving conditions in the allowance areas are generally severe, and the maintenance schedules used reflected that kind of driving. For the DC area, we assumed that driving conditions are normal, and the maintenance schedules used for that area reflected that kind of driving. We combined the recommended frequency of performing each of these jobs with the prices charged by local dealers and service stations to compute an estimated annual maintenance expense. We collected the cost of the complete maintenance service or repair job for each vehicle. For example, we collected the cost of a complete oil change for each vehicle, including the total charge for parts and the total charge for labor. Previous research conducted for OPM revealed varying replacement cycles for constant velocity joint (CVJ) boots among the Alaska allowance areas and between the Alaska areas and the DC area. These were: Anchorage and Juneau--every 45,000 miles (3 years), Nome--every 30,000 miles (2 years), Fairbanks--every 15,000 miles (1 year), and the Washington, DC, area--every 60,000 miles (4 years). We used the Washington, DC, area frequency of repair for the other (i.e., non- Alaska) COLA areas. In each area, we factored the cost of replacement for all three vehicle types into the indexes based upon the frequency of the replacement. In Fairbanks, for example, we included 100 percent of the cost because previous research indicated annual replacement was the norm. To determine the frequency of replacement of windshields, we contacted local dealers and automobile repair shops. Based on the information obtained, we determined that windshield replacement was much more frequent in Alaska than in the other allowance areas or the Washington, DC, area. Therefore, we assumed that windshields had to be replaced every 2 years in the Alaska areas but rarely (i.e., never) in the other areas or in the DC area during the 4-year trade cycle used in the COLA model. The owner's automotive insurance normally covers windshield replacement. Therefore, we used the deductible rather than the surveyed price of windshield replacement, since the deductible was always less than the replacement prices. 5.2.5 Tires Research previously conducted for OPM revealed that various factors (e.g., road quality/state of repair, road composition) appeared to reduce tread life (i.e., the average number of miles a tire is expected to last) in the allowance areas compared with the Washington, DC, area. Based on this research, the model uses tire expense based on a 40,000- mile tread life in allowance areas and a 55,000-mile tread life in the DC area. We priced the cost of a new set of tires, including mounting and balancing and all applicable taxes, in each area. We converted this cost into an annual cost by dividing the estimated number of annual miles driven by the expected tread life and multiplying this by the new tire price. Previous research indicated that four extra studded snow tires would be required for all three vehicles in the Alaska allowance areas (but not in the DC area). Therefore, we surveyed the prices of studded snow tires for all vehicles in Anchorage, Fairbanks, Juneau, and Nome. We also priced the cost of rims and switching snow and street tires semi-annually in these Alaska areas. 5.2.6 License and Registration Fees and Miscellaneous Taxes We obtained information regarding license and registration fees, miscellaneous taxes, and personal property taxes (where applicable). We included license and registration fees as part of the annual cost of owning an automobile. We computed miscellaneous and personal-property taxes for each year of the vehicle's 4-year trade cycle using the vehicle's estimated used-car value for each year. We then averaged the resulting four personal property tax values and included that average as part of the annual cost of owning an automobile. As stated in section 5.2.1, we included sales and excise taxes in the purchase price of the vehicle and accounted for them under the annual vehicle purchase and finance costs. We also include vehicle inspection fees in any area that requires periodic vehicle inspections. 5.2.7 Depreciation The single largest annual expense related to owning and operating a new car is depreciation--the lost value of the vehicle as it ages and is driven. The COLA model calculates total depreciation by subtracting from the purchase price the estimated residual value (used car value) 4 years later. The model then divides this value by four to produce an annual depreciation amount. As described earlier, the new car price was the manufacturer's suggested retail price plus any additional charges, such as shipping, dealer prep, additional dealer markup, documentation fees, excise tax, and sales tax. We based the used car value on information from sources such as the Kelly Blue Book. Although such sources track prices of vehicles sold only in the contiguous 48 States, previous research performed by a contractor for OPM did not indicate that used cars in allowance areas were (on average) worth more or less than used cars in the DC area, except for Fairbanks and Nome. For Fairbanks and Nome, we used 90 percent of the projected residual values to reflect more severe conditions. We note that identical residual values did not result in identical depreciation [[Page 44117]] amounts. Depreciation amounts were generally higher in the allowance areas than in the Washington, DC, area because new car prices were generally higher in the allowance areas. 5.2.8 Finance Expense The COLA model assumes that employees finance new car purchases. Therefore, we surveyed banks in all areas to obtain their auto-loan interest rates for a 48-month loan with 80 percent financing. We computed the finance cost for each vehicle in each area and included it in the annual cost of owning and operating an automobile. 5.2.9 Vehicle Insurance We surveyed the cost of car insurance in each location using the following common coverages, limits, and deductibles: Bodily Injury....................... $100,000/$300,000. Property Damage..................... $25,000. Medical............................. $15,000. Uninsured Motorist.................. $100,000/$300,000. Comprehensive....................... $100 Deductible. Collision........................... $250 Deductible. For the 1998 surveys, we adjusted the limits for Property Damage and Medical based on recommendations from insurance carriers during the 1997 surveys. In each survey area, we identified the common automobile insurance companies and attempted to obtain three insurance price quotes for each type of car surveyed. We averaged these quotes by type of car to produce estimated insurance costs for each area. As in previous surveys, we found that some insurance companies in Guam, Puerto Rico, and the Virgin Islands did not offer the coverages, limits, and deductibles shown above. To allow the comparison of the cost of these different policies with Washington, DC, area costs, we surveyed the cost of insurance in the DC area with comparable offerings in the three allowance areas. We then compared the costs of these equivalent policies to derive adjustment factors that could be applied to the cost of the standard coverage shown above. By applying these factors to the DC area average price, we estimated the cost of equivalent coverage for these particular allowance areas. Appendix 15 shows the factors and their derivation. 5.2.10 Overall Annual Costs As described above, we surveyed the annual costs for fuel, maintenance and oil, tires, licensing, taxes, depreciation, finance, and insurance for three types of automobiles in each allowance area and in the Washington, DC, area. We then summed these costs to determine the overall annual costs by area for owning and operating each type of automobile. Appendix 14 shows these costs for each area by type of vehicle. 5.3 Other Transportation Costs--Air Fares Air fare is the only item we price for the Other Transportation Costs Category. For this item, we surveyed the lowest priced round-trip air fare on a major carrier with a 3-week advance purchase, a 1-week stay over, and travel on Tuesdays and Thursdays. In the previous survey, we used Monday as the travel day. In this survey we used Tuesday (departure date) and Thursday (return date) to avoid peak business travel days and reflect choices consumers might make for recreational travel. While the selection of Tuesday and Thursday as travel days tended to reduce airfares for all areas, it greatly reduced airfares from the Washington, DC, area. This substantially raised the airfare index for each of the COLA areas. We priced trips from each allowance area and the Washington, DC, area to Chicago, Los Angeles, Miami, New York, Seattle, St. Louis, and Omaha. We selected these cities to represent a range of travel destinations coast-to-coast for COLA-area and DC-area Federal employees. To compute the category indexes, we averaged the costs of the trips from each allowance area and then compared these average costs with the average cost of the trips from the DC area. Appendix 16 shows the fares. 5.4 Transportation Component Analyses We compared the total cost of private auto transportation for each vehicle in each allowance area with the total cost for the same vehicle in the Washington, DC, area. We express these comparisons as indexes and show them in Appendix 17. Likewise, we compared the cost of air fares for each area with those for the DC area and computed a cost index. Appendixes 16 and 18 show these indexes. We used national average expenditure data to derive weights that reflected how much consumers typically spend to own and operate an automobile versus other transportation expenses. We used these weights, which vary by income level, to combine the Automobile Expense Category index with the Other Transportation Costs index by area to derive the overall Transportation Component index for the area. Appendix 18 shows the weights, computations, and final Transportation Component indexes. 6. Miscellaneous Expenses 6.1 Component Overview The Miscellaneous Expense Component consists of four categories of expenses: Medical care. Private education (K-12). Contributions (including gifts to non-family members). Personal insurance and retirement contributions/ investments. 6.2 Component Weights We used CES data to determine the appropriate weights for each of the items and categories in the Miscellaneous Expense Component. We show the category weights in Table 6-1 and in Appendix 20. Appendix 19 shows item weights. Table 6-1.--Miscellaneous Expense Categories and Weights ------------------------------------------------------------------------ Income level -------------------------------------- Categories Lower Middle Upper (percent) (percent) (percent) ------------------------------------------------------------------------ Medical care..................... 40.96 31.24 24.27 Private education (K-12)......... 0.98 1.26 1.45 Contributions.................... 16.63 16.27 16.01 Personal insurance and retirement 41.44 51.24 58.27 contributions................... -------------------------------------- Totals....................... 100.00 100.00 100.00 ------------------------------------------------------------------------ Note: Values may not total 100 because of rounding. [[Page 44118]] 6.3 Component Categories 6.3.1 Medical Expense Category We surveyed the price of medical care items using essentially the same approach we used for the Goods and Services Component items. We priced the following medical care items in each allowance area and in the Washington, DC, area: Nonprescription pain reliever Prescription drugs Contact lenses Dental service Doctor visit Hospital room Federal health insurance In addition, we surveyed the price of hospital attendant services in Puerto Rico and air ambulance insurance in the U.S. Virgin Islands. We researched these services during the 1997 surveys, and we found that hospital attendant services were available only in Puerto Rico, where hospital services are significantly different from those in the Washington, DC, area. Therefore, we added the price of hospital attendant service to the price of a hospital room in Puerto Rico. We also found air ambulance insurance to be available only in the Virgin Islands, where on-island hospital services are limited. Therefore, we added the price of air ambulance insurance to the cost of health insurance in the Virgin Islands. We used Federal employee health benefit enrollment information from OPM's Central Personnel Data File along with Federal health benefit premiums to compute average health benefit expense by areas. These expenses varied by area, and we used these averages rather than assuming that costs were constant among areas. We surveyed the cost of the health care items in both the allowance areas and in the DC area. We compared the prices to produce an index for each item in each area, then combined these indexes using CES weights to produce a single Medical Care Category index for each area. 6.3.2 Private Education (K-12) Category Since not everyone sends their children to private school, we derived use factors from the results of the 1992/93 Federal Employee Housing and Living Patterns Survey. Table 6-2 shows these factors and the resulting adjustment of price indexes by area. The factors reflect the relative extent to which Federal employees make use of private education in the COLA areas compared with the Washington, DC, area. For example, the table indicates a use factor of 4.1066 for Puerto Rico because about 54 percent of Federal employees with school age children there send at least one child to private school, compared with about 13 percent for the DC area. Table 6-2.--Summary of Private Education Use Factors and Indexes ---------------------------------------------------------------------------------------------------------------- Employees w/children in private schools Price index Location -------------------------- Use factor Price index w/use Local area DC area factor ---------------------------------------------------------------------------------------------------------------- Anchorage...................................... 10.34 13.23 0.7816 55.53 43.40 Fairbanks...................................... 8.56 13.23 0.6470 41.59 26.91 Juneau......................................... 12.43 13.23 0.9395 57.30 53.84 Nome........................................... 8.08 13.23 0.6107 38.42 23.46 Honolulu....................................... 26.86 13.23 2.0302 113.03 229.48 Hilo\*\........................................ 18.94 13.23 1.4316 44.23 63.32 Kona\*\........................................ 18.94 13.23 1.4316 87.03 124.59 Kauai.......................................... 22.46 13.23 1.6977 95.72 162.50 Maui........................................... 20.39 13.23 1.5412 89.05 137.24 Guam........................................... 42.26 13.23 3.1943 90.95 290.52 Puerto Rico.................................... 54.33 13.23 4.1066 66.85 274.52 St. Croix...................................... 57.27 13.23 4.3288 90.26 390.72 St. Thomas..................................... 51.90 13.23 3.9229 95.78 375.74 ---------------------------------------------------------------------------------------------------------------- * Use data available only for Hawaii County. 6.3.3 Contributions Category The index for the Contributions Category is the same as the Goods and Services Component index for the area. We use the Goods and Services index based on our assumption that the relative level of contributions is roughly equivalent to that reflected by the Goods and Services index. 6.3.4--Personal Insurance and Retirement Category We assume the index for personal insurance and retirement contributions and investments to be constant among areas. The cost of Federal Employees Group Life Insurance is a matter of personal preference and is constant in all areas for the same age, salary, and benefit option combinations. Likewise, retirement contributions are a matter of personal preference, and the minimum contribution requirements are constant among areas for equivalent salary levels. 6.4 Miscellaneous Expense Analyses As with the Goods and Services Component, we combined the indexes for each of the Miscellaneous Component categories using CES weights to produce component indexes by income level for each area. Appendix 20 shows these indexes. Section 2.6 describes how we combine miscellaneous expense component indexes with the other component indexes to derive the final index for each area. 7. Final Results 7.1 Total Comparative Cost Indexes The total comparative cost indexes appear in Table 7-1. Appendix 22 shows how we derived each index from the component indexes. Table 7-1.--Final Cost Comparison Indexes ------------------------------------------------------------------------ Allowance area Index ------------------------------------------------------------------------ Anchorage, Alaska.............................................. 105.65 Fairbanks, Alaska.............................................. 109.19 Juneau, Alaska................................................. 110.46 The rest of the State of Alaska................................ 131.58 City and County of Honolulu, Hawaii............................ 124.51 Hawaii County, Hawaii.......................................... 110.89 Kauai County, Hawaii........................................... 117.19 Maui County, Hawaii............................................ 120.32 Guam/CNMI, Local Retail........................................ 125.23 Guam/CNMI, Commissary/Exchange................................. 121.12 [[Page 44119]] Puerto Rico.................................................... 105.93 U.S. Virgin Islands............................................ 116.33 ------------------------------------------------------------------------ Appendix 1--Publication in the Federal Register of Prior Survey Results: 1990-1998 ------------------------------------------------------------------------ Citation Title Contents ------------------------------------------------------------------------ 56 FR 7902............. Cost-of-Living Results of summer 1990 Allowances and Post living-cost surveys Differentials conducted in Alaska, (Nonforeign Areas). Hawaii, Guam, Puerto Rico, and the U.S. Virgin Islands. 57 FR 58556............ Report on 1991/1992 Results of summer 1991 Surveys Used to and winter 1992 Determine Cost-of- living-cost surveys Living Allowances in conducted in Alaska, Nonforeign Areas. Hawaii, Guam, Puerto Rico, and the U.S. Virgin Islands. 58 FR 45558............ Report on 1992/1993 Results of summer 1992 Surveys Used to and winter 1993 Determine Cost-of- living-cost surveys Living Allowances in conducted in Alaska, Nonforeign Areas. Hawaii, Guam, Puerto Rico, and the U.S. Virgin Islands. 58 FR 27316............ Report on Summer 1993 Results of summer 1993 Surveys Used to living-cost surveys Determine Cost-of- conducted in Hawaii, Living Allowances in Guam, Puerto Rico, Nonforegin areas. and the U.S. Virgin Islands. 59 FR 45066............ Report on Winter 1994 Results of winter 1994 Surveys Used to living-cost surveys Determine Cost-of- conducted in Alaska. Living allowances in Alaska.. 60 FR Report on Summer 1994 Results of summer 1994 6133 Surveys Used to living-cost surveys 2. Determine Cost-of- conducted in Hawaii, Living Allowances in Guam, Puerto Rico, Selected Nonforeign and the U.S. Virgin Areas. Islands. 61 FR Report on Winter 1995 Results of winter 1995 4070. Surveys Used to living-cost surveys Determine Cost-of- conducted in Alaska. Living Allowances in Alaska. 61 FR Report on 1996 Surveys Results of 1996 living- 1419 Used to Determine Cost- cost surveys 0. of-Living Allowances conducted in Alaska, in Nonforeign Areas. Hawaii, Guam, Puerto Rico, and the U.S. Virgin Islands. 63 FR Report on 1997 Surveys Results of 1997 living- 5643 Used to Determine Cost- cost surveys 2. of-Living Allowances conducted in Hawaii, in Nonforeign Areas. Guam, Puerto Rico, and the U.S. Virgin Islands. ------------------------------------------------------------------------ Appendix 2.--Federal Employment Weights Multiple Income Levels: 1998 Survey [Data from multiple income levels within a single allowance area] ---------------------------------------------------------------------------------------------------------------- Location and income level 1995 1996 1998 Average Weights ---------------------------------------------------------------------------------------------------------------- Anchorage: Lower...................................... 1,540 1,445 1,401 1,462 27.02 Middle..................................... 1,754 1,719 1,500 1,658 30.64 Upper...................................... 2,522 2,448 1,903 2,291 42.34 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 5,411 100.00 Fairbanks: Lower...................................... 388 449 466 434 35.20 Middle..................................... 446 456 386 429 34.79 Upper...................................... 405 397 308 370 30.01 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 1,233 100.00 Juneau: Lower...................................... 139 126 100 122 18.91 Middle..................................... 203 199 174 192 29.77 Upper...................................... 341 346 306 331 51.32 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 645 100.00 Rest of Alaska: Lower...................................... 349 363 306 339 23.96 Middle..................................... 703 687 543 644 45.51 Upper...................................... 481 462 352 432 30.53 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 1,415 100.00 Honolulu: Lower...................................... 4,140 4,453 3,919 4,171 33.01 Middle..................................... 3,952 4,009 3,858 3,940 31.19 Upper...................................... 4,514 4,476 4,580 4,523 35.80 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 12,634 100.00 Hawaii: Lower...................................... 139 152 138 143 35.40 Middle..................................... 164 163 160 162 40.10 Upper...................................... 98 101 99 99 24.50 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 404 100.00 Kauai: [[Page 44120]] Lower...................................... 73 59 51 61 27.23 Middle..................................... 76 80 64 73 32.59 Upper...................................... 97 92 80 90 40.18 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 224 100.00 Maui: Lower...................................... 35 35 23 31 22.79 Middle..................................... 59 62 60 60 44.12 Upper...................................... 51 51 33 45 33.09 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 136 100.00 Guam/CNMI: Lower...................................... 947 873 763 861 45.15 Middle..................................... 669 640 561 623 32.67 Upper...................................... 464 430 375 423 22.18 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 1,907 100.00 Puerto Rico: Lower...................................... 2,370 2,281 2,205 2,285 39.89 Middle..................................... 2,166 2,177 2,073 2,139 37.34 Upper...................................... 1,303 1,286 1,322 1,304 22.77 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 5,728 100.00 Virgin Islands: Lower...................................... 98 123 88 103 32.49 Middle..................................... 133 137 130 133 41.96 Upper...................................... 83 76 84 81 25.55 ---------------------------------------------------------------- Totals................................. ........... ........... ........... 317 100.00 ---------------------------------------------------------------------------------------------------------------- Multiple Survey Areas: 1998 Survey [Data from multiple survey areas within a single allowance area] ---------------------------------------------------------------------------------------------------------------- Location 1995 1996 1998 Average Weights ---------------------------------------------------------------------------------------------------------------- Hawaii County: Hilo...................................... 304 308 300 304 75.81 Kona...................................... 97 96 97 97 24.19 ---------------------------------------------------------------- Totals................................ ........... ........... ........... 401 100.00 ================================================================ Virgin Islands: St. Croix................................. 154 166 140 153 48.26 St. Thomas/St. John....................... 160 170 162 164 51.74 ---------------------------------------------------------------- Totals................................ ........... ........... ........... 31 100.00 ---------------------------------------------------------------------------------------------------------------- Appendix 3--Consumer Expenditure Surveys Pre-published Data for All Consumer Units Nationwide* ---------------------------------------------------------------------------------------------------------------- Total complete reporting --------------------------------------------------- 1994 1995 1997 Average ---------------------------------------------------------------------------------------------------------------- Average before tax income.................................. 36,838.00 36,948.00 39,926.00 37,904.00 Average annual expenditures................................ 32,762.99 33,610.38 36,145.95 34,173.11 Food................................................... 4,526.94 4,690.51 4,902.06 4,706.50 Food at home........................................... 2,764.21 2,885.98 2,970.28 2,873.49 Cereals and bakery products........................ 439.36 454.64 464.66 452.89 Cereals and cereal products........................ 166.94 169.16 165.56 167.22 Flour.............................................. 7.93 8.93 8.94 8.60 Prepared flour mixes............................... 13.20 13.29 16.51 14.33 Ready-to-eat and cooked cereals.................... 102.02 99.83 92.76 98.20 Rice............................................... 15.47 19.43 18.21 17.70 Pasta, cornmeal and other cereal products.......... 28.32 27.68 29.13 28.38 Bakery products.................................... 272.42 285.49 299.10 285.67 Bread.............................................. 77.20 78.18 86.16 80.51 White bread........................................ 38.02 38.37 42.35 39.58 Bread, other than white............................ 39.17 39.81 43.81 40.93 [[Page 44121]] Crackers and cookies............................... 64.36 70.09 70.06 68.17 Cookies............................................ 43.78 46.76 45.86 45.47 Crackers........................................... 20.58 23.33 24.19 22.70 Frozen and refrigerated bakery products............ 22.16 22.42 23.43 22.67 Other bakery products.............................. 108.70 114.79 119.45 114.31 Biscuits and rolls................................. 37.26