Poverty Measurement: Adjusting for Geographic Cost-of-Living Difference
(Letter Report, 03/09/95, GAO/GGD-95-64).

This report provides information on the statistical data requirements
that would be needed to construct a cost-of-living index that could be
used, at the federal level, to adjust for geographic differences in
living costs.  Concerns had been raised in Congress that current
measures do not recognize that residents of high-cost areas may need
higher incomes to meet their basic needs.  GAO (1) describes the
function of market baskets in determining a cost-of-living index,
including both a uniform national market basket and market baskets that
reflect regional differences in consumption; (2) identifies
methodologies that might be used to calculate a cost-of-living
adjustment, including methodologies that researchers and private
industry use for comparing costs by geographic areas; and (3) presents
expert opinions on the ability of these methodologies to adjust the
poverty measurements for geographic differences in cost of living.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-95-64
     TITLE:  Poverty Measurement: Adjusting for Geographic 
             Cost-of-Living Difference
      DATE:  03/09/95
   SUBJECT:  Cost of living
             Disadvantaged persons
             Price indexes
             Demographic data
             Income statistics
             Economic analysis
             Statistical methods
             Economic research
IDENTIFIER:  USDA Survey of Food Consumption
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to Congressional Requesters

March 1995

POVERTY MEASUREMENT - ADJUSTING
FOR GEOGRAPHIC COST-OF-LIVING
DIFFERENCE

GAO/GGD-95-64

Poverty Measurement

(243056)


Abbreviations
=============================================================== ABBREV

  BLS - Bureau of Labor Statistics
  COL - Cost of living
  USDA - Department of Agriculture

Letter
=============================================================== LETTER


B-259782

March 9, 1995

The Honorable James M.  Jeffords
The Honorable Joseph I.  Lieberman
The Honorable Daniel Patrick Moynihan
United States Senate

Because of your concern that the determination of poverty status does
not reflect geographic differences in the cost of living (COL), you
asked us to provide information about the statistical data
requirements that would be needed to construct a COL index that could
be used, at the federal level, to adjust for geographic differences
in living costs.  The current measurement to determine poverty levels
does not take into account geographic differences in COL.  You were
concerned that the current measure does not adequately recognize that
residents of high-cost areas may need higher incomes to adequately
meet their basic needs. 

Specifically, we agreed to (1) describe the function of market
baskets in determining a COL index,\1 including both a uniform
national market basket and market baskets that reflect regional
differences in consumption; (2) identify methodologies that might
have potential for calculating a COL adjustment, including
methodologies that researchers and private industry use for comparing
costs by geographic areas; and (3) obtain expert opinions on the
ability of these methodologies to adjust the poverty measurement for
geographic differences in COL.  As we agreed with your offices, our
work was limited to the technical feasibility of these methodologies. 
We did not seek to obtain and verify data on the cost of the
methodologies, nor did we evaluate whether it would be appropriate to
incorporate a COL adjustment into benefit formulas for public
assistance programs for the poor.\2


--------------------
\1 A market basket is a listing of goods and services that is deemed
to represent a particular economic standing or well-being.  For
example, a market basket used to measure the changes in prices in
urban areas comprises all the goods and services consumed by urban
households.  Goods in that market basket stay the same from one
period to the next.  The items in a market basket for a COL index,
however, change to reflect a constant standard of living from one
period to another.  A COL index is calculated with price information
on the items in a market basket over time.  A market basket for
poverty determination comprises items defined as those that would
constitute a reasonable social minimum for the population. 

\2 At the time when we were doing our work, a panel of the National
Academy of Sciences was conducting a broader study on concepts,
information needs, and measurement methods for poverty and family
assistance.  Among its tasks, the panel was to assess and make
recommendations on methodologies to adjust for inflation over time
and for geographical differences.  The panel's report was to be
issued by September 1994, but it has been delayed.  Therefore, we
could not evaluate any recommendations that this panel may make on
geographic adjustments for COL. 


   BACKGROUND
------------------------------------------------------------ Letter :1

In 1969, the federal government officially adopted a measure to
ascertain how many people across the country had incomes that were
inadequate to meet expenses for basic needs.  This poverty measure
was based on the finding of the U.S.  Department of Agriculture's
(USDA) 1955 Survey of Food Consumption that, on average, families of
three or more persons spent one-third of their income on food. 
Poverty for a family of three was computed as three times the cost of
the economy food plan, the least costly food plan designed by USDA.\3
The poverty measure has been updated annually with a COL index to
adjust for the change in prices nationwide, but the poverty measure
has not been adjusted for differences in prices by geographic area. 
Thus, in 1993, a family of three with a cash income of less than
$11,522 was considered to be living in poverty, regardless of place
of residence. 

The concept of geographic COL adjustments of poverty measurement has
been seen as problematic.  A 1976 report to Congress on the
measurement of poverty stated that "one of the most troublesome
concepts of poverty measurement" was making adjustments for
geographic differences in COL.\4 It ultimately concluded that
unresolved conceptual issues, such as the development of generally
accepted market baskets of goods and services representative of the
needs of the poor in various geographic areas, and data limitations
precluded satisfactory geographic adjustments.  More recently, in a
1992 report, we noted that there was insufficient data on which to
base geographic adjustments to the measure of poverty.\5

Some economists contend that adjusting the poverty measure for
geographic differences in COL would be inappropriate, irrespective of
the methodology used.  They say that any such adjustment to reflect
regional differences in market baskets would fail to recognize other
regional differences that are relevant to a definition of poverty or
the needs of the poor.  For example, a COL index probably would not
reflect differences among geographic areas in the level of support or
assistance available to low-income families. 


--------------------
\3 The original measure was created by adjusting for such factors as
family size, sex and age of the family head, number of children under
18 years old, and farm/nonfarm residence.  In 1981, distinctions
based on sex of the family head and farm/nonfarm residence were
eliminated, and additional changes were made for families of nine of
more members. 

\4 U.S.  Department of Health, Education, and Welfare, The Measure of
Poverty:  A Report to Congress as Mandated by the Education
Amendments of 1974 (Washington, D.C.:  U.S.  Government Printing
Office, Apr.  1976), pp.  81-82. 

\5 See Poverty Trends, 1980-1988:  Changes in Family Composition and
Income Sources Among the Poor (GAO/PEMD-92-34, Sept.  10, 1992). 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :2

Market baskets provide the foundation for any measure of COL. 
Obtaining a consensus on what should go into market baskets for a COL
index and on how to keep them current would be difficult.  Even if
consensus were obtained on the specific items and their quantities to
include in market baskets for a COL index, problems would arise in
identifying market baskets that reflect a constant standard of
living. 

We identified 12 methodologies that, in some part, could contribute
to an index that potentially could be used to adjust poverty
measurement to reflect geographical differences in COL.  Some of the
methodologies rely on different ways of defining market baskets that
could reflect geographic differences, and others employ approaches
for adjusting the prices of goods and services in previously defined
market baskets.  A few of the methodologies both define market
baskets and adjust the prices of the items to derive a COL index. 
Additionally, a few of the methodologies are now used by private
industry and the federal government to adjust wages and salaries for
geographic differences in COL.  Others are solely conceptual
methodologies and are not used for such adjustment. 

In the collective view of the experts we asked to assess these
methodologies, the long-standing problems involved in identifying a
method to adjust poverty measurement for geographic differences in
COL have not been resolved; data and conceptual problems have
prevented any adjustment in the past and continue to do so today. 
Overall, experts' ratings of the methodologies were mixed.  Experts'
comments about each methodology's strengths and weaknesses were
diverse, and sometimes conflicting. 


   SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :3

To address our first two objectives, describing the function of a
market basket and identifying potential methods for calculating a COL
adjustment, we reviewed the relevant literature on measuring poverty
and on geographic adjustment for COL and discussed these issues with
specialists.  These specialists included individuals associated with
poverty measurement or COL data at the Bureau of Labor Statistics
(BLS) and the Bureau of the Census, as well as private organizations
and academic institutions.  On the basis of these reviews and
discussions, we identified 12 methodologies that might have potential
for adjusting poverty measures to reflect geographic differences in
COL.  We consider these 12 methodologies to be illustrative for a
wide range of potential approaches to determine geographic COL
differences, but recognize that the list is not, and cannot be,
exhaustive.  (A more detailed account of our scope and methodology is
contained in app.  I.)

To meet our third objective of obtaining expert opinion on the
ability of the methodologies to adjust the poverty measure for
geographic differences in COL, we identified experts and asked them
to review the methodologies.  From our list of more than 40 potential
experts compiled during our literature review and initial discussions
with specialists, we selected 15 experts to review the methodologies. 
(See app.  II for a list of the selected experts.)

We sent a questionnaire to these experts in which we described each
methodology briefly.  We asked the experts to review each of the 12
methodologies and to categorize the methodology's potential for use
in adjusting the poverty measurement for geographic difference in
COL.  Additionally, we asked them to discuss the strengths and
weaknesses of each methodology.  (See app.  III for a copy of the
information and questionnaire sent to each expert.) All 15 experts
responded and we tabulated their ratings for each methodology to
determine the ones the experts considered most and least promising. 
We also analyzed the written responses on strengths and weaknesses. 

We did our work in Washington, D.C., between September 1994 and
January 1995 in accordance with generally accepted government
auditing standards.  Because we did not evaluate the policies or
operations of any federal agency to develop the information presented
in this report, we did not seek comments from any agency. 


   MARKET BASKETS ARE NECESSARY,
   BUT A UNIFORM NATIONAL MARKET
   BASKET IS NEITHER USED NOR
   CONSIDERED DESIRABLE BY EXPERTS
------------------------------------------------------------ Letter :4

Market baskets of goods and services form the basis for determining a
COL index.  Of the methodologies we examined that calculate a COL
index, none uses a uniform national market basket in which the same
quantities of identical goods and services are used in all locations. 
In fact, these methodologies all used market baskets that have
different measures for at least one component--for example,
transportation or housing.  Several of the experts, in their comments
on COL methodologies, said that market baskets for COL indexes should
vary to reflect differences in local standards of living. 


      MARKET BASKETS ARE THE
      FOUNDATION OF A COL INDEX
---------------------------------------------------------- Letter :4.1

Market baskets of goods and services provide the foundation for
determining COL.  The composition of the market baskets, such as the
items included or the quantity of one item included in relation to
other items, affects the dollar values that are determined to
represent COL.  Conceptually, market baskets for a COL index would
accurately reflect differences in tastes, as well as needs, such that
an individual would derive equal satisfaction from the various market
baskets priced in different geographic locations.  For example, food
preferences in southeastern states for low-cost cereals, such as rice
and corn, lowers COL in these areas, while climatic differences
necessitates the expenditures for heating a home and warm clothing
and increases the COL in northern states. 

Obtaining a consensus on what should go into a COL index's market
baskets and on how to update them would be difficult.  The method
generally preferred by the experts we contacted to determine the
items to include in market baskets is to use expert judgment to
specify the requirements for physical health and social well-being. 
But standards have not been identified for the majority of components
of a COL index's market baskets.  Even if consensus were obtained on
the specific items and their quantities to include in a COL index's
market baskets, another problem would be how to keep the market
baskets up to date to reflect a constant standard of living. 


      A UNIFORM NATIONAL MARKET
      BASKET IS NOT USED IN COL
      INDEXES
---------------------------------------------------------- Letter :4.2

Of the methodologies we examined that calculate a COL index, all used
market baskets that reflected regional differences in standards of
needs and/or actual consumption patterns.\6 Most notably, these
methodologies varied in how they determined the housing and
transportation components of the market baskets by adjusting for
regional variation. 


--------------------
\6 Some of the 12 methodologies that we identified did not actually
calculate a COL index, but provided regional estimates of costs with
which we calculated regional ratios.  These ratios were used to
demonstrate geographic differences that might be obtained with the
methodologies.  In this part of our report on market baskets, we
examined only those methodologies that calculated a COL index using a
market basket.  Those with regional ratios, which did not have market
baskets, are excluded in this discussion of market baskets. 


      EXPERTS SAY THAT LOCAL
      STANDARDS OF LIVING ARE
      NECESSARY IN A COL INDEX'S
      MARKET BASKETS
---------------------------------------------------------- Letter :4.3

We received numerous comments about market baskets for a COL index
from the experts from whom we solicited assessments of the
methodologies.  Several experts noted the need to adjust the
composition of the market baskets for differences in local standards
of living among geographic areas.  One expert commented that it is
nearly impossible to obtain reliable evidence or credible expert
judgments about the composition of market baskets to reflect specific
local standards of living.  This expert suggested that market baskets
should be changed as acceptable standards are developed.  The problem
of keeping market baskets up to date was noted by other experts in
their comments about the use of outdated data and concepts.  For
example, one expert specifically wanted a child care component to be
included in the market baskets. 


   METHODOLOGIES EXIST THAT
   POTENTIALLY COULD LEAD TO A
   GEOGRAPHIC COL INDEX
------------------------------------------------------------ Letter :5

We identified 12 generic methodologies that, in some part, could
contribute to the development of a COL index that potentially could
be used to adjust the poverty measurement for geographic differences. 
Four methodologies identified baseline data,\7 or developed a market
basket that could be the basis for constructing a COL index by
geographic area.  Six methodologies calculated a COL index from
existing cost data or a previously defined market basket.  Two
methodologies developed an original market basket, collected data,
and calculated a COL index with those data.  Table 1 provides
descriptions of the 12 methodologies.  (Detailed descriptions of
these methodologies are found in app.  III.)



                                Table 1
                
                  Brief Descriptions of Methodologies

Methodology             Description
----------------------  ----------------------------------------------
Budgets                 Estimates how much families need to spend to
                        purchase the contents of a market basket of
                        goods and services.

Norms                   Using existing data and specifically collected
                        data, calculates COL indexes for specific
                        geographic locations by using standards
                        developed by identifying the proportion of
                        income spent on consumer expenditure
                        categories.

Housing data            Estimates average rents of housing in
                        particular geographic areas.

Family budgets          Using data collected to measure inflation,
                        calculates the annual estimates of the cost of
                        purchasing hypothetical market baskets of
                        goods and services that represent lower,
                        intermediate, and higher standards of living.

Consumption data        Calculates the average dollar amount of what
                        families report that they spend in specified
                        expenditure categories, such as clothing,
                        during the period of data collection.

Interarea price index   Develops price index numbers from BLS' pricing
                        and item-characteristics information, assumes
                        equal consumer satisfaction among geographic
                        areas, and allows a direct comparison of
                        relative prices among several geographic
                        areas.

Economic modeling       Develops COL indexes for specific geographic
                        locations by using information published by
                        government agencies and private sources.

Consumer price index    Determines COL index for specific geographic
                        areas by applying annual average price changes
                        to baseline data.\

Estimation models       Empirically determines COL indexes for
                        specific geographic areas (such as regions,
                        states, and counties) from baseline data by
                        using statistical procedures.

Local indexes           Using data specifically collected from local
                        price surveys, develops a COL index.

Polling                 Uses public opinion survey data to establish a
                        measure of the level of income that people
                        think should represent the poverty line for a
                        family in the respondents' specific community.

Comparable pay          Calculates employers' costs per hour worked
                        for each of the components of labor
                        compensation--wages and salaries and employee
                        benefits.
----------------------------------------------------------------------
A few of the methodologies are now used as COL indexes, but most have
not been.  For example, the norms, local indexes, and economic
modeling methodologies are used in the private sector as COL indexes
to make geographic COL adjustments for pay and relocation decisions. 
Until their discontinuance in 1981, estimates from the family budgets
methodology had been used by policymakers to set income eligibility
criteria for employment programs and to geographically adjust wages
and salaries.  Several of the methodologies that identify baseline
data are used in ways other than to show differences in COL.  For
example, USDA uses the consumption data methodology to estimate
expenditures on a child, which then are used to determine payments
for the support of children in foster families.  Many of the
methodologies were developed by researchers to develop indexes to
reflect COL differences, such as those categorized under the
estimation models, interarea price index, and the consumer price
index methodologies; but none of these are used to make geographic
COL adjustments.  (See app.  III for detailed descriptions of how the
data and indexes from the 12 methodologies are used.)

We identified two additional methodologies but could not locate
research that delineated how the methodologies could be implemented
to develop a COL index.  For example, administrative data from public
assistance programs, such as the food stamp program, have been
proposed as baseline data for developing a COL adjustment that would
indicate the incidence of need within a geographic location. 
However, in our review of the relevant literature and discussions
with specialists, we did not locate appropriate data that could be
translated into an index to demonstrate geographic variation. 
Another approach to identify baseline data for a COL index would be
to use information obtained from grocery stores' universal product
code scanners.  As in the case of administrative program data, we
could not locate information that indicated how the product code data
could be used to develop a geographic index or ratio. 

During the process of obtaining experts' ratings of promise for the
12 generic methodologies we identified, some experts indicated that
we had not identified and presented all possible methodologies to
make such a COL adjustment.  A number of the experts suggested using
a combination of several attributes from the methodologies that they
reviewed.  In addition, they identified four other methodologies that
could be considered for doing geographic COL adjustments.  One was a
modification of the local indexes methodology, and another was a
modeling technique to develop regional variables to obtain baseline
data.  The other two focused on ways to revise the current poverty
measurement.  One methodology included the most basic levels of
shelter and food as the basis for measuring poverty.  The other
methodology, according to an expert, is what the National Academy of
Sciences panel is expected to recommend in its forthcoming report. 
None of these methodologies was identified by more than one of the
experts, however. 

We recognize that our list of 12 methodologies is not exhaustive, but
consider it to provide a fair overview of the wide range of
alternatives.  The fact that the experts suggested further
methodologies, and that no alternative was proposed by more than one
expert, suggests that no agreement now exists among experts as to the
best way to adjust the measurement of poverty for geographic
differences in COL.  This is discussed in the next section. 


--------------------
\7 Baseline data refer to costs or estimates that could be used to
calculate a COL index.  These data might identify the cost of several
items in a market basket, or of a single item, such as the cost of
renting an apartment in the housing data methodology described in
appendix III.  Baseline data need not include all of the goods and
items that comprise a COL market basket. 


   EXPERTS DIFFERED ON THE
   METHODOLOGIES' ABILITY TO MAKE
   ADJUSTMENTS
------------------------------------------------------------ Letter :6

The observation in a 1976 report to Congress that "although there may
be geographic differences in the cost of living, there is no known
way to make satisfactory geographic adjustments to the poverty
cutoffs,"\8 still seems valid.  The experts who we asked to assess
the methodologies differed about how best to make adjustments because
of numerous data and conceptual problems that they identified. 
Overall, the experts' ratings of each methodology's promise for
geographically adjusting COL were mixed, and our content analysis of
the experts' comments about each methodology's strengths and
weaknesses yielded diverse and sometimes conflicting perspectives. 


--------------------
\8 U.S.  Department of Health, Education, and Welfare, The Measure of
Poverty:  A Report to Congress as Mandated by the Education
Amendments of 1974, p.  82. 


      EXPERTS VIEWED METHODOLOGIES
      AS HAVING MIXED DEGREES OF
      PROMISE
---------------------------------------------------------- Letter :6.1

Overall, the experts' ratings of methodologies were mixed.  Although
the majority of experts rated certain methodologies as showing little
or no promise for adjusting the poverty measurement for geographic
differences in COL, no clear consensus was observed overall in the
ratings the experts gave regarding the methodologies' promise for
making adjustments.  A majority of the experts regarded local
indexes, polling, family budgets, consumption data, and the consumer
price index methodologies as showing little or no promise for making
adjustments.  The comparable pay methodology was found by more than
two-thirds of the experts to be not promising at all.  (See table 2
for experts' ratings of methodologies.)



                                     Table 2
                     
                        Experts' Ratings of Methodologies

                        Very          Modera             Not
                       great   Great      te  Little  promis  Cannot say
                      promis  promis  promis  promis  ing at     at this
Methodology                e       e       e       e     all        time   Total
--------------------  ------  ------  ------  ------  ------  ----------  ------
Budgets                    1       5       6       1       2           0      15
Norms                      0       4       5       4       1           1      15
Housing data               1       2       5       4       3           0      15
Family budgets             3       2       2       5       3           0      15
Consumption data           1       3       3       5       3           0      15
Interarea price            1       2       4       7       0           1      15
 index
Economic modeling          1       2       4       3       1           4      15
Consumer price index       0       2       5       5       3           0      15
Estimation models          0       1       5       4       3           2      15
Local indexes              0       1       4       9       1           0      15
Polling                    1       1       1       3       7           1    14\a
Comparable pay             0       0       2       2      11           0      15
--------------------------------------------------------------------------------
\a In one instance, the response was not usable. 

Source:  GAO analysis. 

No methodology was rated by the majority of experts as showing great
or very great promise to adjust the poverty measurement for
geographic differences in COL.  However, three
methodologies--budgets, norms, and housing data--received a rating of
at least moderate promise by a majority of the experts.  The budgets
methodology appeared to have the most promise, but less than half of
the experts rated it as having great or very great promise. 


      EXPERTS IDENTIFIED STRENGTHS
      AND WEAKNESSES FOR PROMISING
      METHODOLOGIES
---------------------------------------------------------- Letter :6.2

Our content analysis of the experts' comments on each methodology's
strengths and weaknesses showed that the experts shared few common
views on any specific methodology.  When three or more experts did
express a similar comment, it most often concerned a weakness rather
than a strength of the methodology being rated.  Some experts
identified an attribute but expressed different perspectives as to
whether it constituted a strength or weakness.  Examples of mixed
responses included one expert indicating that a strength of a
particular methodology was its adaptability for use by government,
while another expert characterized the same methodology as not being
adaptable for use by government.  In some instances, experts agreed
about a methodology's attribute--e.g., its emphasis on children--but
differed as to whether the presence of this attribute should be
viewed as a strength or weakness.  (See figure 1 for strengths and
weaknesses of the methodologies.)

   Figure 1:  Experts' Comments on
   Strengths and Weaknesses of
   Methodologies

   (See figure in printed
   edition.)

Note:  At least three experts had to comment on a methodology's
attribute for it to be included in this table.  For mixed responses,
at least one expert cited an attribute as a strength, while another
or others rated it as a weakness. 

Source:  GAO tallies of expert reviewers' comments on strengths and
weaknesses of methodologies. 



   (See figure in printed
   edition.)

Our content analysis of the experts' comments on the strengths and
weaknesses of the three methodologies that received a rating of at
least moderate promise by the majority of experts illustrates both
the diverse and occasionally contradictory comments of the experts. 
The strengths of the budgets methodology lie in its representation of
low-income families and its use of health and social well-being
standards in the determination of the market basket.  However, its
eclectic approach of using these standards from various sources,
which makes it difficult to explain to laypersons, was viewed as a
weakness.  Another weakness of the budgets methodology cited by the
experts is that it fails to make adjustments for regional differences
in transportation and some of the other market basket components. 
The experts who commented about its use of expenditure data were
evenly split between those who viewed this as a strength and those
who said it was a weakness.  This methodology was viewed as capturing
both contemporary and outdated concepts of consumption needs.  For
example, one expert cited the use of current standards as a strength,
whereas other experts cited the use of 1981-based data to determine
the importance given to items in the market basket as a weakness. 

The norms methodology was generally rated as promising because the
COL index was frequently updated.  The experts, however, differed in
their comments about the methodology.  For example, more than
one-half of the experts said that the lowest income level for which
the index was provided was well above poverty and was therefore
unrepresentative of low-income families.  Conversely, one expert,
noting the degree of variation in income levels provided in the
index, described it as "more relevant to the poor than other
available sources." Mixed responses of both strengths and weaknesses
were indicated for the (1) appropriateness of the items in the market
basket, (2) degree of geographic variation shown in the index, (3)
ability of the methodology to be adapted and implemented by the
government, and (4) cost associated with such implementation. 

The housing data methodology was regarded as strong in its focus on
what the experts considered the major source of variation in COL. 
The fact that housing was the only cost measured was also cited as
this methodology's major weakness.  As shown in table 3, the experts
had mixed views about the representation in the baseline data of
families living in poverty.  The experts also lacked agreement on
whether the housing concepts were appropriate.  For example, one
expert said the methodology had the "merit of focusing on rents for a
specified type of apartment," while another said that "decent, safe,
and sanitary" qualities of housing should be controlled in the
measure to prevent downward bias in low-income areas. 


      WEAKNESSES IDENTIFIED FOR
      METHODOLOGIES WITH LITTLE OR
      NO PROMISE
---------------------------------------------------------- Letter :6.3

A content analysis of the experts' comments revealed that the local
indexes methodology had many weaknesses resulting from its price data
collection methods, which involve volunteers from chambers of
commerce collecting and averaging prices that are representative of
purchases of middle-management households in their local areas.  This
methodology was viewed to be an unsuitable representation of the
consumption needs of the poor.  Another weakness of the local indexes
methodology was its exclusion of nonmetropolitan and rural areas. 

The polling methodology was regarded by several experts as a means to
validate the measurement of poverty, rather than as an approach to
make geographic COL adjustments.  These experts said that this
methodology provided insight into the relationship between an
absolute measure of poverty, such as the current official measure,
and a measure that is relative--that is, a measure that changes with
growth in the economy or according to society's perception of an
adequate level of income.  According to the experts' comments, the
main weakness of polling was in the quality of the data obtained
through a public opinion survey.  It was thought that the respondents
would be biased in providing their estimates.  For example, one
expert wrote:  "If respondents knew the survey results would be used
to adjust poverty thresholds with implications for program
expenditures and income taxes, then some may intentionally deflate or
inflate their response[s], in their own self-interest." The experts
had mixed views about the costs associated with this method; some
experts said it would be cost effective, while others said it would
be costly. 

According to the experts' comments, the main weakness of the
comparable pay methodology was its reliance on employers' labor
costs.  Many experts said that such a measure included influences
other than COL and that as a consequence it was inappropriate and an
unsuitable substitute for COL, especially as a representation of the
needs of the poor.  For example, one expert said, "Geographic
variations in quality of life affect the relationship between
wages/salaries and living costs.  Use of employer costs as a measure
of living costs would introduce significant regional bias." Many
weaknesses, as well as several mixed responses, were noted for the
remaining three methodologies--consumption data, family budgets, and
consumer price index. 


   CONCLUSION
------------------------------------------------------------ Letter :7

The concept of adjusting the measurement of poverty for geographic
differences in COL has been seen as problematic, and remains so.  We
asked recognized experts to review 12 methodologies that illustrate
the range of alternative approaches to adjust poverty measurement for
geographic COL differences, and there was no consensus among these
experts that any one methodology was the most promising for making
such an adjustment.  The fact that several of these experts suggested
additional methodologies, but that no additional methodology was
suggested by more than one of the experts, suggests to us that a
consensus on any one approach does not exist.  Where there does
appear to be agreement, however, is that several of the methodologies
offer little or no promise of appropriately adjusting the measurement
of poverty for geographic COL differences.  Further, obtaining a
consensus on what items should go into a COL index's market baskets
to reflect regional differences in consumption would be difficult. 


---------------------------------------------------------- Letter :7.1

As arranged with your offices, unless you publicly announce its
contents earlier, we plan no further distribution of this report
until 20 days after its issue date.  At that time, we will send
copies of the report to the Secretary of Commerce, the Secretary of
Labor, the Director of the Office of Management and Budget, and other
interested parties.  We will also make copies available to others on
request. 

If you have any questions concerning this report, please call me on
(202) 512-8676.  Major contributors to the report are listed in
appendix IV. 

William M.  Hunt
Director, Federal
 Management Issues


SCOPE AND METHODOLOGY
=========================================================== Appendix I

To address the first two objectives of this job--describing the
function of a market basket in determining a COL index, and
identifying potential methods for calculating a COL adjustment--we
first reviewed the relevant literature and held discussions with
specialists in the field.  These specialists included individuals
associated with poverty measurement or COL data at the Bureau of
Labor Statistics (BLS) and the Bureau of the Census, as well as
private organizations and academic institutions.  We also included
individuals who did not support geographic adjustment of the poverty
measurement, as well as those who have proposed methodologies to
achieve this objective. 

On the basis of our literature review and preliminary discussions
with specialists, we described the function of a market basket and
identified an initial set of methodologies that might have potential
for adjusting poverty measurement for geographic differences in the
COL.  We grouped similar methodologies into 12 categories and gave a
generic name to each.  We excluded potential methodologies if they
did not identify existing data that could be turned into a
geographically adjusted index.  Two methods, one based on use of data
from administrative records and one relying on data scanning of
uniform product codes, were eliminated because they did not meet this
criterion. 

To meet our third objective of obtaining expert opinion on the
ability of these methodologies to adjust the poverty measure for
geographic differences in COL, we selected a panel of 15 experts and
surveyed them using a data collection instrument that contained brief
descriptions of each of the 12 generic methodologies we identified. 
We asked the panel to review each description and rate each
methodology in terms of its promise for use in adjusting the poverty
measurement for geographic differences in COL. 

The description of each methodology identified data sources,
discussed the cost and time needed to develop an index with the
methodology, and provided an example of how the calculations would be
made and the index could be used.  We asked the developer or someone
very familiar with each methodology to review our brief description
to ensure that it accurately conveyed the essence of the methodology. 

We asked the selected experts to rate each methodology on a
five-point scale that ranged from "not promising at all" to "shows
very great promise," and then briefly discuss the strengths and
weaknesses of the methodology.  The experts were also asked to
identify any additional methodology we may have overlooked and
provide their views on the major challenges and costs associated with
developing COL data that could be used to geographically adjust the
poverty measure. 

We randomly chose 15 individuals to serve as experts from a candidate
list of more than 40 names.  To obtain a diverse candidate pool
reflective of the different interests involved, we asked for
nominations of potential experts from those specialists in the field
and representatives of major statistical agencies that we met with
during our initial discussions and literature review.  To avoid
potential conflicts of interest, we excluded individuals from the
list who are currently serving on the National Academy of Sciences'
Panel on Poverty and Family Assistance or who are political
appointees.  We recognize that the responses we received reflect only
the views of the experts included. 

Several of the experts initially selected were unable to participate. 
We replaced these individuals with alternates from the remaining pool
of candidates.  (See app.  II for a list of the participating
experts.)

Before contacting our initial selections, we asked congressional
staff and officials from Census, BLS, and the Office of Management
and Budget to review the list for balance and to identify any
additional experts they believed should be included.  No additions
were suggested. 

The selected experts received a package containing a letter of
introduction, an instruction sheet, descriptions of all the
methodologies, and response sheets (see app.  III).  The package was
sent on November 14, 1994. 

Responses were received from all 15 experts by January 6, 1995.  We
tabulated the ratings for each methodology to obtain an overall
assessment of the experts' opinions of how promising each methodology
was for use in adjusting the poverty threshold for geographic
differences in COL. 

We also did a content analysis of the experts' responses to the
strengths and weaknesses question for each methodology.  From an
initial reading of the responses, we developed a list of cited
strengths and weaknesses.  We used this list to code the responses of
all experts for each methodology.  The coding of the responses was
verified by a second coder, and a third person checked coding
reliability.  As a method of focusing our analysis on the recurring
comments made by the experts in their discussions of each
methodology's strengths and weaknesses, we adopted a decision rule to
report only those comments made by three or more experts for a
particular methodology's attribute. 

Experts' comments on market baskets were identified separately and
were used in our description of the function of the market basket. 
Additionally, we used experts' general comments on major challenges
and costs associated with geographically adjusting poverty measures
to illustrate our results. 


EXPERTS AND THEIR AFFILIATIONS
========================================================== Appendix II

Mark C.  Berger
University of Kentucky

Dixie Blackley
Le Moyne College

Tom Carlin
Department of Agriculture

David Cutler
Harvard University

Anne Draper
AFL-CIO

Lawrence Gibson
Eric Marder Associates, Inc. 

Robert Gillingham
Department of Treasury

Haeduck Lee
The World Bank

Richard Muth
Emory University

Marilyn Moon
The Urban Institute

Tom R.  Rex
Arizona State University

Patricia Ruggles
Joint Economic Committee

Timothy M.  Smeeding
Syracuse University

Robert Summers
University of Pennsylvania

Harold Watts
Columbia University


DATA COLLECTION INSTRUMENT AND
DESCRIPTIONS OF THE METHODOLOGIES
========================================================= Appendix III

This appendix contains copies of the cover letter, instruction sheet,
answer sheets, and brief descriptions of the 12 methodologies that we
sent to the 15 experts we selected to review the methodologies. 



(See figure in printed edition.)COVER LETTER



(See figure in printed edition.)INSTRUCTION SHEET



(See figure in printed edition.)BUDGETS METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)NORMS METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)HOUSING DATA METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)FAMILY BUDGETS METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)CONSUMPTION DATA METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)INTERAREA PRICE INDEX METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)ECONOMIC MODELING METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)CONSUMER PRICE INDEX METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)ESTIMATION MODELS METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)LOCAL INDEXES METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)POLLING METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)COMPARABLE PAY METHODOLOGY



(See figure in printed edition.)



(See figure in printed edition.)ADDITIONAL METHODOLOGIES



(See figure in printed edition.)SUMMARY QUESTION



(See figure in printed edition.)ADDITIONAL COMMENTS


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================== Appendix IV

Kathleen K.  Scholl, Senior Economist
Thomas M.  Beall
Timothy J.  Carr
James M.  McDermott
Patrick R.  Mullen
Pamela R.  Pavord
A.  Elizabeth Powell
Kiki Theodoropoulos

RELATED GAO PRODUCTS

Federal Aid:  Revising Poverty Statistics Affects Fairness of
Allocation Formulas (GAO/HEHS-94-165, May 20, 1994). 

Poverty Trends, 1980-88:  Changes in Family Composition and Income
Sources Among the Poor (GAO/PEMD-92-34, Sept.  10, 1992). 


*** End of document. ***