THIS IS A REMINDER THAT REBECCA BLANK'S NEXT POVERTY MEASUREMENT MEETING
WILL TAKE PLACE ON TUESDAY, MAY 26, 3:30-5:00PM, IN THE OLD EXECUTIVE
OFFICE BUILDING, ROOM 324.
THIS MEETING IS A DEPUTIES DECISION MEETING, AS DESCRIBED ON THE ATTACHED
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND. Unable to convert ARMS_EXT: [ATTACH.D13jMAIL47180693J.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000B0100002800C8196810480D000011090000005AOOOB010000103600540069006D0065
00730020004E0065007700200052006F006D0061006E00200052006500670075006C0061007200
000000000000000001000200580201000000040028000000000000000000000000000000000000
00011202002400A1000000A10000000A000000400301000200410303000200420302005B004303
010002004403020002004503010045004603010002004703010044004803010002004903010002
00D37D04230000000000000000000000000000000008337C007800000200004903000003010004
0002000000DDOA10008301040003000200211000DDDDOBOB00030000040BOODD9BD2000E000002
OOC003B0040EOOD2D2010E00000200C003B0040EOOD2D1000EOOOO0200C003B0040EOOD1D1010E
00000200C003B0040EOOD1F20CF2CCF30CF3E0300C000000001806OCOOEOEOOOOCOOOOOOOOC003
OCOOEOE0301400000000180618061824180618241400EOEOOOOCOOOOOOOOC0030COOEOE0401AOO
000000A80AA80A1100EC1318061824180618241AOOEOF20CF2506F7665727479804D6561737572
656D656E7480506F6C69637980576F726B696E67809C47726F7570F30CF3808484804D61798032
3688D0041500000B00090001B004F000010001201500DOE0401200OOOOOOEC13EC132AOOEC1312
OOE088CC546865807072696D61727980707572706F7365806F668074686973806D656574696E67
80697380746F8074727980616E6480726561636880636F6E73656E7375738061726F756E648061
806B657980736574806F6680706F6C696379CF7265636F6D6D656E646174696F6E738072656C61
74656480746F8074686580697373756573806F75728067726F757080686173806265656E806469
7363757373696E67806F766572807468658070617374807468726565806D6F6E7468732E80CF54
6865736580617265807265636F6D6D656E646174696F6E7380746F807468658043656E73757380
Hex-Dump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall CPI or by a cpr for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a cpr for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at all) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
. that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the cpr for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
Hex-Dump Conversion
NAS Recommendations on How to Measure Poverty
A poverty measure consists of two pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family of four (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
. 30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other farnily sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK)F
where A is the number of adults and K the number of kids. P is the scaling for children (if P= 1, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs ofliving are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the ~xpenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Hex-Dump Conversion
Policy Issue 1: Determining the Threshold
Agenda
1_ How does the NAS propose to detennine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indetenninancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Hex-Dump Conversion
Policy Issue 2: Updating the Poverty Thresholds Over Time
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. Ifwe don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards of living, it's much harder to interpret the poverty statistics.
2. Standard of living changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate t~ese would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives. .
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
Hex-Dump Conversion
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresMlds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons ofthese approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account of MOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account ofMOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND. ==================== ATTACHMENT
ATT CREATION TIME/DATE: 0 00:00:00.00
TEXT:
Unable to convert ARMS_EXT: [ATTACH.D13jMAIL47180693J.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000B0100002800C8196810480D000011090000005AOOOB010000103600540069006D0065
00730020004E0065007700200052006F006D0061006E00200052006500670075006C0061007200
000000000000000001000200580201000000040028000000000000000000000000000000000000
00011202002400A1000000A10000000A000000400301000200410303000200420302005B004303
010002004403020002004503010045004603010002004703010044004803010002004903010002
00D37D04230000000000000000000000000000000008337C007800000200004903000003010004
0002000000DDOA10008301040003000200211000DDDDOBOB00030000040BOODD9BD2000E000002
OOC003B0040EOOD2D2010E00000200C003B0040EOOD2D1000EOOOO0200C003B0040EOOD1D1010E
00000200C003B0040EOOD1F20CF2CCF30CF3E0300C000000001806OCOOEOEOOOOCOOOOOOOOC003
OCOOEOE0301400000000180618061824180618241400EOEOOOOCOOOOOOOOC0030COOEOE0401AOO
000000A80AA80A1100EC1318061824180618241AOOEOF20CF2506F7665727479804D6561737572
656D656E7480506F6C69637980576F726B696E67809C47726F7570F30CF3808484804D61798032
3688D0041500000B00090001B004F000010001201500DOE0401200OOOOOOEC13EC132AOOEC1312
OOE088CC546865807072696D61727980707572706F7365806F668074686973806D656574696E67
80697380746F8074727980616E6480726561636880636F6E73656E7375738061726F756E648061
806B657980736574806F6680706F6C696379CF7265636F6D6D656E646174696F6E738072656C61
74656480746F8074686580697373756573806F75728067726F757080686173806265656E806469
7363757373696E67806F766572807468658070617374807468726565806D6F6E7468732E80CF54
6865736580617265807265636F6D6D656E646174696F6E7380746F807468658043656E73757380
Hex-Dump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall CPI or by a cpr for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a cpr for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at all) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
. that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the cpr for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
Hex-Dump Conversion
NAS Recommendations on How to Measure Poverty
A poverty measure consists of two pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family of four (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
. 30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other farnily sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK)F
where A is the number of adults and K the number of kids. P is the scaling for children (if P= 1, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs ofliving are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the ~xpenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Hex-Dump Conversion
Policy Issue 1: Determining the Threshold
Agenda
1_ How does the NAS propose to detennine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indetenninancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Hex-Dump Conversion
Policy Issue 2: Updating the Poverty Thresholds Over Time
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. Ifwe don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards of living, it's much harder to interpret the poverty statistics.
2. Standard of living changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate t~ese would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives. .
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
Hex-Dump Conversion
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresMlds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons ofthese approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account of MOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account ofMOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
THIS IS A REMINDER THAT REBECCA BLANK'S NEXT POVERTY MEASUREMENT MEETING
WILL TAKE PLACE ON TUESDAY, MAY 26, 3:30-5:00PM, IN THE OLD EXECUTIVE
OFFICE BUILDING, ROOM 324.
THIS MEETING IS A DEPUTIES DECISION MEETING, AS DESCRIBED ON THE ATTACHED
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND. Unable to convert ARMS_EXT: [ATTACH.D13]MAIL47180693J.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOQOOOOOOOOOOOOOO000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000B0100002800C8196810480D000011090000005AOOOB010000103600540069006D0065
00730020004E0065007700200052006F006D0061006E00200052006500670075006C0061007200
000000000000000001000200580201000000040028000000000000000000000000000000000000
00011202002400A1000000A10000000A000000400301000200410303000200420302005B004303
010002004403020002004503010045004603010002004703010044004803010002004903010002
00D37D04230000000000000000000000000000000008337C007800000200004903000003010004
0002000000DDOA10008301040003000200211000DDDDOBOB00030000040BOODD9BD2000E000002
00C003B0040EOOD2D2010E00000200C003B0040EOOD2D1000EOOOO0200C003B0040EOOD1DIOIOE
00000200C003B0040EOODIF20CF2CCF30CF3E0300C000000001806OCOOEOEOOOOCOOOOOOOOC003
OCOOEOE0301400000000180618061824180618241400EOEOOOOCOO000000C0030COOEOE0401AOO
000000A80AA80AIIOOEC13L8061824180618241AOOEOF20CF2506F7665727479804D6561737572
656D656E7480506F6C69637980576F726B696E67809C47726F7570F30CF3808484804D61798032
3688D0041500000B00090001B004FOOOOI0001201500DOE0401200000000EC13EC132AOOEC1312
00E088CC546865807072696D61727980707572706F7365806F668074686973806D656574696E67
80697380746F8074727980616E6480726561636880636F6E73656E7375738061726F756E648061
806B657980736574806F6680706F6C696379CF7265636F6D6D656E646174696F6E738072656C61
74656480746F8074686580697373756573806F75728067726F757080686173806265656E806469
7363757373696E67806F766572807468658070617374807468726565806D6F6E7468732E80CF54
6865736580617265807265636F6D6D656E646174696F6E7380746F807468658043656E73757380
L1 eCOrdsM,
ryex-DlJrn anagern,
'P ConVers' ent 8YSle
Poverty Measurement Policy Working Group -- May 26 IOn rn
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary ofthe NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall cpr or by a cpr for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a cpr for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at aU) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreernent.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the cpr for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
A poverty measure consists of two pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family of four (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PKl
where A is the number of adults and K the number of kids. P is the scaling for children (ifP=I, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs ofliving are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Agenda
I. How does the NAS propose to detennine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indetenninancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms ofthese different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards ofliving change, so do social
norms.
2. If we don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level ofrnedian expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
I. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards ofliving, it's much harder to interpret the poverty statistics.
2. Standard ofliving changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
, "18/eel
Ii, "'ecorOj
Agenda 'i9.t{),,1JJ. 'S 4181}8"
If) Cal} "'i9"'el}/
vel3'lO/J 8YSfe",
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
'Il7 !edR,
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures liek~:~%anage""
Onve"' en! S/f~'
'';10" ""Iel!}
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons of these approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems of lack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account ofMOOP in the future will
involve relatively small changes to the poverty numbers. If we don't do it now and want to take
account of MOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
,....
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND. ==================== ATTACHMENT
ATT CREATION TIME/DATE: 0 00:00:00.00
TEXT:
Unable to convert ARMS_EXT: [ATTACH.D13]MAIL47180693J.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOQOOOOOOOOOOOOOO000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000B0100002800C8196810480D000011090000005AOOOB010000103600540069006D0065
00730020004E0065007700200052006F006D0061006E00200052006500670075006C0061007200
000000000000000001000200580201000000040028000000000000000000000000000000000000
00011202002400A1000000A10000000A000000400301000200410303000200420302005B004303
010002004403020002004503010045004603010002004703010044004803010002004903010002
00D37D04230000000000000000000000000000000008337C007800000200004903000003010004
0002000000DDOA10008301040003000200211000DDDDOBOB00030000040BOODD9BD2000E000002
00C003B0040EOOD2D2010E00000200C003B0040EOOD2D1000EOOOO0200C003B0040EOOD1DIOIOE
00000200C003B0040EOODIF20CF2CCF30CF3E0300C000000001806OCOOEOEOOOOCOOOOOOOOC003
OCOOEOE0301400000000180618061824180618241400EOEOOOOCOO000000C0030COOEOE0401AOO
000000A80AA80AIIOOEC13L8061824180618241AOOEOF20CF2506F7665727479804D6561737572
656D656E7480506F6C69637980576F726B696E67809C47726F7570F30CF3808484804D61798032
3688D0041500000B00090001B004FOOOOI0001201500DOE0401200000000EC13EC132AOOEC1312
00E088CC546865807072696D61727980707572706F7365806F668074686973806D656574696E67
80697380746F8074727980616E6480726561636880636F6E73656E7375738061726F756E648061
806B657980736574806F6680706F6C696379CF7265636F6D6D656E646174696F6E738072656C61
74656480746F8074686580697373756573806F75728067726F757080686173806265656E806469
7363757373696E67806F766572807468658070617374807468726565806D6F6E7468732E80CF54
6865736580617265807265636F6D6D656E646174696F6E7380746F807468658043656E73757380
L1 eCOrdsM,
ryex-DlJrn anagern,
'P ConVers' ent 8YSle
Poverty Measurement Policy Working Group -- May 26 IOn rn
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary ofthe NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall cpr or by a cpr for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a cpr for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at aU) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreernent.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the cpr for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
A poverty measure consists of two pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family of four (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PKl
where A is the number of adults and K the number of kids. P is the scaling for children (ifP=I, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs ofliving are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Agenda
I. How does the NAS propose to detennine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indetenninancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms ofthese different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards ofliving change, so do social
norms.
2. If we don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level ofrnedian expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
I. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards ofliving, it's much harder to interpret the poverty statistics.
2. Standard ofliving changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
, "18/eel
Ii, "'ecorOj
Agenda 'i9.t{),,1JJ. 'S 4181}8"
If) Cal} "'i9"'el}/
vel3'lO/J 8YSfe",
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
'Il7 !edR,
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures liek~:~%anage""
Onve"' en! S/f~'
'';10" ""Iel!}
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons of these approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems of lack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account ofMOOP in the future will
involve relatively small changes to the poverty numbers. If we don't do it now and want to take
account of MOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
,....
I think they really want you there. To be honest, most of the issues
they've raised have not struck me as things that have huge, obvious policy
implications, but I have not been able to devotes much time to this.
Becky and Pat Ruggles, along with folks at OMB, and to a lesser extent
NEC, have given a lot of thought to the issues. I will go back through my
notes and send you a note by Monday with any of the major issues flagged.
I think they'd like to have you there if at all possible though I told
them it was a tough day for me with our big welfare to work event the next
day.
---------------------- Forwarded by Andrea Kane/OPD/EOP on OS/22/98 08:24
AM ---------------------------
Francine P. Obermiller
OS/21/98 04:50:00 PM
Record Type: Record
To: See the distribution list at the bottom of this message
cc:
Subject: MAY 26-POVERTY MEASUREMENT MEETING
THIS IS A REMINDER THAT REBECCA BLANK'S NEXT POVERTY MEASUREMENT MEETING
WILL TAKE PLACE ON TUESDAY, MAY 26, 3:30-5:00PM, IN THE OLD EXECUTIVE
OFFICE BUILDING, ROOM 324.
THIS MEETING IS A DEPUTIES DECISION MEETING, AS DESCRIBED ON THE ATTACHED
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND.
Message Sent
TO:~~------~--------~~~--~----------------------------------
patricia ruggles @ 690-6562 @ fax
Mark A. Wasserman/OMB/EOP
Richard B. Bavier/OMB/EOP
Joseph J. Minarik/OMB/EOP
Ophelia D. West/OMB/EOP
Sally Katzen/OPD/EOP
Cecilia E. Rouse/OPD/EOP
Donna L. Geisbert/OPD/EOP
Cynthia A. Rice/OPD/EOP
Andrea Kane/OPD/EOP
Rebecca M. Blank/CEA/EOP
Katherine K. Wallman/OMB/EOP
Paul Bugg/OMB/EOP
Anil Kakani/OMB/EOP
Shannon Mason/OPD/EOP
Francine P. Obermiller/CEA/EOP
Donald R. Arbuckle/OMB/EOP
Lisa M. Jones/OMB/EOP
Elena Kagan/OPD/EOP
Shannon Mason/OPD/EOP
Unable to convert ARMS_EXT: [ATTACH.D49]MAIL448313148.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
.7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000B0100002800C8196810480D000011090000005AOOOBOI0000103600540069006D0065
00730020004E0065007700200052006F006D0061006E00200052006500670075006C0061007200
000000000000000001000200580201000000040028000000000000000000000000000000000000
00011202002400A1000000A10000000A000000400301000200410303000200420302005B004303
010002004403020002004503010045004603010002004703010044004803010002004903010002
00D37D04230000000000000000000000000000000008337C007800000200004903000003010004
0002000000DDOA10008301040003000200211000DDDDOBOB00030000040BOODD9BD2000E000002
00C003B0040EOOD2D2010E00000200C003B0040EOOD2D1000EOOOO0200C003B0040EOOD1D1010E
00000200C003B0040EOOD1F20CF2CCF30CF3E0300C000000001806OCOOEOEOOOOCOOOOOOOOC003
OCOOEOE0301400000000180618061824180618241400EOEOOOOCOO000000C0030COOEOE0401AOO
000000A80AA80A1100EC1318061824180618241AOOEOF20CF2506F7665727479804D6561737572
656D656E7480506F6C69637980576F726B696E67809C47726F7570F30CF3808484804D61798032
3688D0041500000B00090001B004F000010001201500DOE0401200000000EC13EC132AOOEC1312
00E088CC546865807072696D61727980707572706F7365806F668074686973806D656574696E67
80697380746F8074727980616E6480726561636880636F6E73656E7375738061726F756E648061
806B657980736574806F6680706F6C696379CF7265636F6D6D656E646174696F6E738072656C61
74656480746F8074686580697373756573806F75728067726F757080686173806265656E806469
7363757373696E67806F766572807468658070617374807468726565806D6F6E7468732E80CF54
6865736580617265807265636F6D6D656E646174696F6E7380746F807468658043656E73757380
61626F757480686F778074686579806D696768748070726573656E7480746865697280616C7465
Hex-Dump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall CPI or by a CPI for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a CPI for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at all) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
. (Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the CPI for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
Hex-Dump Conversion
NAS Recommendations on How to Measure Poverty
A poverty measure consists of two pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs. (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family of four (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK/
where A is the number of adults and K the number of kids. P is the scaling for children (if P= 1, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs ofliving are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Hex-Dump Conversion
Policy Issue 1: Determining the Threshold
Agenda
1. How does the NAS propose to determine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indeterminancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
*******~**************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Hex-Dump Conversion
Policy Issue 2: Updating the Poverty Thresholds Over Time
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments'
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. If we don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities. .
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards ofliving, it's much harder to interpret the poverty statistics.
2. Standard ofliving changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Hex-Dump Conversion
Policy Issue 3: Adjusting for Geographic Variations in Cost of Living
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the.poverty guidelines used for program
purposes, these should not include geographical price variation.
Hex-Dump ConverslOf'
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons of these approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account ofMOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account ofMOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.) .
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
==================== ATTACHMENT 1 ====================
ATT CREATION TIME/DATE: 0 00:00:00.00
TEXT:
Unable to convert ARMS_EXT: [ATTACH.D49]MAIL448313148.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
.7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000B0100002800C8196810480D000011090000005AOOOBOI0000103600540069006D0065
00730020004E0065007700200052006F006D0061006E00200052006500670075006C0061007200
000000000000000001000200580201000000040028000000000000000000000000000000000000
00011202002400A1000000A10000000A000000400301000200410303000200420302005B004303
010002004403020002004503010045004603010002004703010044004803010002004903010002
00D37D04230000000000000000000000000000000008337C007800000200004903000003010004
0002000000DDOA10008301040003000200211000DDDDOBOB00030000040BOODD9BD2000E000002
00C003B0040EOOD2D2010E00000200C003B0040EOOD2D1000EOOOO0200C003B0040EOOD1D1010E
00000200C003B0040EOOD1F20CF2CCF30CF3E0300C000000001806OCOOEOEOOOOCOOOOOOOOC003
OCOOEOE0301400000000180618061824180618241400EOEOOOOCOO000000C0030COOEOE0401AOO
000000A80AA80A1100EC1318061824180618241AOOEOF20CF2506F7665727479804D6561737572
656D656E7480506F6C69637980576F726B696E67809C47726F7570F30CF3808484804D61798032
3688D0041500000B00090001B004F000010001201500DOE0401200000000EC13EC132AOOEC1312
00E088CC546865807072696D61727980707572706F7365806F668074686973806D656574696E67
80697380746F8074727980616E6480726561636880636F6E73656E7375738061726F756E648061
806B657980736574806F6680706F6C696379CF7265636F6D6D656E646174696F6E738072656C61
74656480746F8074686580697373756573806F75728067726F757080686173806265656E806469
7363757373696E67806F766572807468658070617374807468726565806D6F6E7468732E80CF54
6865736580617265807265636F6D6D656E646174696F6E7380746F807468658043656E73757380
61626F757480686F778074686579806D696768748070726573656E7480746865697280616C7465
Hex-Dump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall CPI or by a CPI for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a CPI for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at all) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
. (Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the CPI for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
Hex-Dump Conversion
NAS Recommendations on How to Measure Poverty
A poverty measure consists of two pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs. (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family of four (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK/
where A is the number of adults and K the number of kids. P is the scaling for children (if P= 1, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs ofliving are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Hex-Dump Conversion
Policy Issue 1: Determining the Threshold
Agenda
1. How does the NAS propose to determine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indeterminancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
*******~**************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Hex-Dump Conversion
Policy Issue 2: Updating the Poverty Thresholds Over Time
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments'
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. If we don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities. .
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards ofliving, it's much harder to interpret the poverty statistics.
2. Standard ofliving changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Hex-Dump Conversion
Policy Issue 3: Adjusting for Geographic Variations in Cost of Living
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the.poverty guidelines used for program
purposes, these should not include geographical price variation.
Hex-Dump ConverslOf'
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons of these approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account ofMOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account ofMOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.) .
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
After looking at Becky's summary of the issues (attached), I don't think
there's much else to add. Do you plan to attend?
---------------------- Forwarded by Andrea Kane/OPD/EOP on OS/26/98 08:22
AM ---------------------------
Andrea Kane
OS/22/98 08:25:02 AM
Record Type: Record
To: Elena Kagan/OPD/EOP
cc:
Subject: MAY 26-POVERTY MEASUREMENT MEETING
I think they really want you there. To be honest, most of the issues
they've raised have not struck me as things that have huge, obvious policy
implications, but I have not been able to devotes much time to this.
Becky and Pat Ruggles, along with folks at OMB, and to a lesser extent
NEC, have given a lot of thought to the issues. I will go back through my
notes and send you a note by Monday with any of the major issues flagged.
I think they'd like to have you there if at all possible though I told
them it was a tough day for me with our big welfare to work event the next
day.
---------------------- Forwarded by Andrea Kane/OPD/EOP on OS/22/98 08:24
AM ---------------------------
Francine P. Obermiller
OS/21/98 04:50:00 PM
Record TyPe: Record
To: See the distribution list at the bottom of this message
cc:
Subject: MAY 26-POVERTY MEASUREMENT MEETING
THIS IS A REMINDER THAT REBECCA BLANK'S NEXT POVERTY MEASUREMENT MEETING
WILL TAKE PLACE ON TUESDAY, MAY 26, 3:30-5:00PM, IN THE OLD EXECUTIVE
OFFICE BUILDING, ROOM 324.
THIS MEETING IS A DEPUTIES DECISION MEETING, AS DESCRIBED ON THE ATTACHED
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND.
Message Sent
TO:~ ________________ ~ __________________________________________
patricia ruggles @ 690~6562 @ fax
Mark A. Wasserman/OMB/EOP
Richard B. Bavier/OMB/EOP
Joseph J. Minarik/OMB/EOP
Ophelia D. West/OMB/EOP
Sally Katzen/OPD/EOP
Cecilia E. Rouse/OPD/EOP
Donna L. Geisbert/OPD/EOP
Cynthia A. Rice/OPD/EOP
Andrea Kane/OPD/EOP
Rebecca M. Blank/CEA/EOP
Katherine K. Wallman/OMB/EOP
Paul Bugg/OMB/EOP
Anil Kakani/OMB/EOP
Shannon Mason/OPD/EOP
Francine P. Obermiller/CEA/EOP
Donald R. Arbuckle/OMB/EOP
Lisa M. Jones/OMB/EOP
Elena Kagan/OPD/EOP
Shannon Mason/OPD/EOP
Unable to convert ARMS_EXT: [ATTACH.D46)MAIL45680354B.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722FIF2DOFBEOBB6EI038556D6412EC2C8E244F72342
Hex-Dump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (I) should they be
adjusted on an annual basis by the overall CPI or by a CPI for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a CPI for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at all) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the CPI for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
A poverty measure consists oftwo pieces: (1) A definition offamily resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Defining Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Defining a Poverty Threshold
The threshold should be determined for a family offour (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK)F
where A is the number of adults and K the number of kids. P is the scaling for children (if P= 1, kids
are treated the same as adults); and F is the scaling for family size .. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs of living are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Agenda
1. How does the NAS propose to determine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indeterminancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do. this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the cur.rent poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. Ifwe don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards of living, it's much harder to interpret the poverty statistics.
2. Standard of living changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences acr,oss regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons ofthese approaches?
**********************************************
Pros of handling MOOP as.recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account of MOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account ofMOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* If we ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
==================== ATTACHMENT 1 ====================
ATT CREATION TIME/DATE: 0 00:00:00.00
TEXT:
Unable to convert ARMS_EXT: [ATTACH.D46)MAIL45680354B.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722FIF2DOFBEOBB6EI038556D6412EC2C8E244F72342
Hex-Dump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a general consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate to the 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (I) should they be
adjusted on an annual basis by the overall CPI or by a CPI for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a CPI for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at all) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the CPI for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
A poverty measure consists oftwo pieces: (1) A definition offamily resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Defining Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Defining a Poverty Threshold
The threshold should be determined for a family offour (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK)F
where A is the number of adults and K the number of kids. P is the scaling for children (if P= 1, kids
are treated the same as adults); and F is the scaling for family size .. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs of living are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Agenda
1. How does the NAS propose to determine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indeterminancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do. this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the cur.rent poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations of the NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share of different groups (although all subgroup poverty rates would go up.)
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. Ifwe don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way, i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards of living, it's much harder to interpret the poverty statistics.
2. Standard of living changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences acr,oss regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to much greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons ofthese approaches?
**********************************************
Pros of handling MOOP as.recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account of MOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account ofMOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* If we ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather than subtracting it from income, and would like to see this implemented.
After looking at Becky's summary of the issues (attached), I don't think
there's much else to add. Do you plan to attend?
---------------------- Forwarded by Andrea Kane/OPD/EOP on OS/26/98 08:22
AM ---------------------------
Andrea Kane
OS/22/98 08:25:02 AM
Record Type: Record
To: Elena Kagan/OPD/EOP
cc:
Subject: MAY 26-POVERTY MEASUREMENT MEETING
I think they really want you there. To be honest, most of the issues
they've raised have not struck me as things that have huge, obvious policy
implications, but I have not been able to devotes much time to this.
Becky and Pat Ruggles, along with folks at OMB, and to a lesser extent
NEC, have given a lot of thought to the issues. I will go back through my
notes and send you a note by Monday with any of the major issues flagged.
I think they'd like to have you there if at all possible though I told
them it was a tough day for me with our big welfare to work event the next
day.
---------------------- Forwarded by Andrea Kane/OPD/EOP on OS/22/98 08:24
AM ---------------------------
Francine P. Obermiller
OS/21/98 04:50:00 PM
Record Type: Record
To: See the distribution list at the bottom of this message
cc:
Subject: MAY 26-POVERTY MEASUREMENT MEETING
THIS IS A REMINDER THAT REBECCA BLANK'S NEXT POVERTY MEASUREMENT MEETING
WILL TAKE PLACE ON TUESDAY, MAY 26, 3.:30-5:00PM, IN THE OLD EXECUTIVE
OFFICE BUILDING, ROOM 324.
THIS MEETING IS A DEPUTIES DECISION MEETING, AS DESCRIBED ON THE ATTACHED
AGENDA, , AND DR. BLANK URGES EVERYONE TO ATTEND.
Message Sent
TO:~~ ______~____~~~~__~_________________________________
patricia ruggles @ 690-6562 @ fax
Mark A. Wasserman/OMB/EOP
Richard B. Bavier/OMB/EOP
Joseph J. Minarik/OMB/EOP
Ophelia D. West/OMB/EOP
Sally Katzen/OPD/EOP
Cecilia E. Rouse/OPD/EOP
Donna L. Geisbert/OPD/EOP
Cynthia A. Rice/OPD/EOP
Andrea Kane/OPD/EOP
Rebecca M. Blank/CEA/EOP
Katherine K. Wallman/OMB/EOP
Paul Bugg/OMB/EOP
Anil Kakani/OMB/EOP
Shannon Mason/OPD/EOP
Francine P. Obermiller/CEA/EOP
Donald R. Arbuckle/OMB/EOP
Lisa M. Jones/OMB/EOP
Elena Kagan/OPD/EOP
Shannon Mason/OPD/EOP
Unable to convert ARMS_EXT: [ATTACH.D46]MAIL45680354B.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
HexDump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a generai consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate tothe 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall cpr or by a cpr for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a cpr for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at aU) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the cpr for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
Hex-Dump Conversion
NAS Recommendations on How to Measure Poverty
A poverty measure consists oftwo pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family offour (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK/
where A is the number of adults and K the number of kids. P is the scaling for children (ifP=l, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs of living are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Policy Issue 1: Determining the Threshold Hex-Dump Conversion
Agenda
1. How does the NAS propose to determine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indeterminancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations ofthe NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share ofdifferent groups (although all subgroup poverty rates would go up.)
Hex-Dump Conversion
Policy Issue 2: Updating the Poverty Thresholds Over Time
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. Ifwe don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way,i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards ofliving, it's much harder to interpret the poverty statistics.
2. Standard ofliving changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Hex-Dump Conversion
Policy Issue 3: Adjusting for Geographic Variations in Cost of Living
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to rriuch greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
Hex-Dump Convers;o~
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons of these approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account ofMOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account of MOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather *an subtracting it from income, and would like to see this implemented.
==================== ATTACHMENT 1 ====================
ATT CREATION TIME/DATE: 0 00:00:00.00
TEXT:
Unable to convert ARMS_EXT: [ATTACH.D46]MAIL45680354B.126 to ASCII,
The following is a HEX DUMP:
FF57504370040000010A020100000002050000001841000000020000AF6DF55FOOC4AE4D8EOD78
5BB87D2075CD97A622513B745FC80056A2722F1F2DOFBEOBB6E1038556D6412EC2C8E244F72342
B2EB4C490945B990D4A8F992A6CA9FOE66DBFE1BFOCC576A1437EFC5A8D903F8DA0987COEACE19
B07C8ADOA6FFAC4DED828FF98E13E74FD421E34C72E63142DDCE083621AD0646453A4AC48797A5
D2BD5BEDEE65A2AA6D3033F822288B03E848B28A9154E61BAD01A8EDC779066D90548D2F466099
D25E1817F80FAAC2488CFFC4FA6270C68FA9C540BF214ECD6BA79BC57DF7609EFA94E20F078BDE
7EACCC62760FE4E892B1E87A6F4C9FE4F5A8FC8D7038DODA4D959E75165662F24EA733336B623D
78556783E88EF5018433D2D7776DDE41BE91B5EOAD056BA354BB4B2ABBAA583537C012EDAA6CB5
98CB5E894566CBA6AOCE88719B419EE6C4207BEB3D7EBAC18A138B41DB11A96DBEFCD29147CEOF
7BC782ECFF796ED0122C68D2BD47806E4F3ABOF8A53B4992DFBAACF6F5BD7085F23E7905FE5749
4FB60FA59FD5DF01566B60BE9FC508A8D1659F81CC8B125C597978657EBDA7E35E2B70AOE35A37
EB7711BBBD3B1ECC508FA12880FFA398B9D0548A186761DDA80FODE2B756E3A684F3F40C51546A
2FB81BA8B1922EBOC9C096E5025AC04C07BOAF457EF8E6A191B2EC8328199457C9C736CEC888FB
AE5B650EF802000900000000000000000000000823010000000B0100007E020000005501000000
4E0000008903000009250100000006000000D70300000B300200000028000000DD030000087701
000000400000000504000008340100000014000000450400000802010000000F00000059040000
080501000000080000006804000000984C006F00630061006C0020004800500020004C00610073
0065006A0065007400200035004DOOOOOOOOOOOOOOOOOOOOOOOOOO000000000000000000000000
00000000000000000000000057494E53504F4F4COOOOOOOOOOC800C8002C012C012C012C01C800
C80030000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000000000000000000000000000000000000000000000000000000000000000
HexDump Conversion
Poverty Measurement Policy Working Group -- May 26
The primary purpose of this meeting is to try and reach consensus around a key set of policy
recommendations related to the issues our group has been discussing over the past three months.
These are recommendations to the Census about how they might present their alternative poverty
measurement calculations. Attached to this agenda is a one-page summary of the NAS
recommendations for a new poverty line measure, and a one-page summary of the four policy issues
we've discussed at our last four meetings.
AGENDA
I. Key issues to be discussed
A. How should the thresholds be determined in the first year an alternative measure is published?
To be specific: should the alternative poverty rate be benchmarked in 1997 to be identical to the 1997
official poverty rate or is there an alternative "benchmark" that makes more sense?
(Note: We appear to have a generai consensus within the group from our earlier discussions to support
benchmarking the alternative poverty rate tothe 1997 official rate.)
B. How should the poverty thresholds be updated over time? Specifically, (1) should they be
adjusted on an annual basis by the overall cpr or by a cpr for food shelter & clothing only and
revisited every 5-10 years for a more complete recalculation; or (2) Should they be completely
recalculated each year as a share of current expenditures on food, clothing, and housing?
(Note: There seemed to be general consensus for the first approach, particularly if a cpr for food,
shelter and clothing was used. OMB seemed reasonably confident that they and Census could
establish a process to assure that poverty measurement is reviewed on a regular basis.)
C. Should the poverty thresholds be adjusted for geographic variations in the cost of living?
(Note: There seemed to be general consensus that any "base" alternative poverty measurement should
not include geographical price variation, although Census may well want to publish an alternative that
includes this as part of the NAS recommendations.)
D. How (if at aU) should the adjustment for Medical out-of-pocket (MOOP) expenditures
occur? Specifically, should the Census follow the NAS recommendations to impute MOOP and
subtract it from individual income before calculating individual poverty status?
(Note: There is more disagreement on this issue than on any other. The three deputies present at our
last meeting generally agreed that the Census should follow the NAS recommendations for now, but
that further work refining these calculations (and considering other alternatives) should go forward.
But some of the staff present at the meeting were in less agreement.)
E. Do we have any recommendations to Census about how to present their alternative poverty
calculations? Specifically, do we want to recommend they present a primary "base" alternative
poverty calculation (for instance: benchmarked to 1997, updated by the cpr for necessities, without
geographic price variation, and including MOOP), and then a few alternatives (with and without
MOOP, with and without geographic price variation)?
II. Process from here.
Hex-Dump Conversion
NAS Recommendations on How to Measure Poverty
A poverty measure consists oftwo pieces: (1) A definition of family resources, and (2) A poverty
threshold against which resources are compared. If resources are below the threshold, a family is
considered poor.
(1) Derming Family Resources
Family resources = Cash income
+ Near money in-kind benefits (food stamps, housing subsidies, school lunch, LIHEAP, etc)
- Taxes
- Child care costs (for families in which there is no nonworking parent, with a cap, and not to
exceed the earnings of the parent with lower earnings)
- Work expenses (a flat amount per week of work)
- Child support payments
- Out of pocket medical care expenditures, including health insurance premiums.
(2) Derming a Poverty Threshold
The threshold should be determined for a family offour (two adults and two children) based on a point
in the distribution of annual expenditures by such families on food, clothing, and shelter, plus a small
multiplier. (This is based on Consumer Expenditure Survey data, probably using a three-year average
to guarantee adequate sample size.) The recommended percentage of annual expenditures is between
30 and 35 percent (i.e., 15 to 20 percentage points below the median); recommended multipliers are
between 1.15 to 1.25.
Thresholds for other family sizes should be determined based on an equivalence scale calculation.
The NAS panel recommends using
Scale value = (A + PK/
where A is the number of adults and K the number of kids. P is the scaling for children (ifP=l, kids
are treated the same as adults); and F is the scaling for family size. The NAS panel recommends
P=0.70 and F in a range of 0.65 to 0.75.
Geographic costs of living are taking into account, based on a regional index of housing costs (which
varies by size of metropolitan area and region of country) which is weighted by the share of housing
costs in the expenditure bundle.
Over time, new poverty threshold should be updated by recalculating annual expenditures on the most
recent three years of data, and reapplying these other calculations using the most recent data available
on housing costs.
Policy Issue 1: Determining the Threshold Hex-Dump Conversion
Agenda
1. How does the NAS propose to determine thresholds?
How does this compare to the current poverty line?
2. Comments on the fundamental indeterminancy of the "right" threshold ...
Implies a range of thresholds are viable.
3. Benchmarking a threshold so poverty numbers in a specific year are unchanged.
A. How do this?
B. Why do this?
C. What are the implications of benchmarking?
a. Change history
b. Change who is poor, even if the aggregate number is unchanged
4. What are the alternatives to benchmarking?
NAS recommendations
5. What are the criticisms of these different approaches?
**********************************
Pros of benchmarking:
There may be large political advantages to saying "We want to improve the measure of poverty, but to
show you that we're doing this in a non-political manner, we'll benchmark so there is no change in the
overall aggregate poverty rate in the year we implement this change." There are still those who will be
upset about the changes in the distribution of poverty or in the past history of poverty rates, which will
change (you can benchmark one year only.)
Cons of benchmarking
Backing out a threshold so that the current poverty rate is unchanged produces a threshold that is very
far down in the percentile distribution of expenditures on food, shelter and clothing -- somewhere
around 20-25% rather than the 30-35% range recommended by the NAS.
Pros of using the NAS alternative (which establishes a higher threshold and higher poverty rates)
Utilizes the recommendations ofthe NAS panel, based on their judgement about the accumulation of
evidence.
Maintains threshold levels that are very similar to the current thresholds (it's not clear how much this
matters given the resource definition has changed.)
Cons of using the NAS alternative
Results in a substantially higher poverty rate, which will draw an great deal of criticism. At the same
time, it has the same problems as benchmarking, in that it also changes history and it changes the
relative poverty share ofdifferent groups (although all subgroup poverty rates would go up.)
Hex-Dump Conversion
Policy Issue 2: Updating the Poverty Thresholds Over Time
Agenda
1. How have the poverty thresholds been updated?
CPI adjustments
2. What does the NAS recommend?
Recalculating the threshold
3. What are the implications of these different approaches?
Absolute poverty thresholds (current approach)
Relatively-absolutely poverty thresholds (NAS approach)
Relative poverty thresholds
4. What are the alternatives?
A. Adjust with CPI
B. Adjust with CPI but embed a commitment to recalculate thresholds
on a regular basis.
C. Utilize the relatively-absolute (NAS) technique annually
5. What are the criticisms of these different approaches?
*********************************************
Pros of a more relative approach (B or C above):
1. CPI-adjusted thresholds become obsolete over time. As standards of living change, so do social
norms.
2. Ifwe don't create the expectation of a regular recalculation/update of the poverty line, we'll end up
in 30 years exactly where we are today, using an outmoded statistic.
3. Ifwe calculate the thresholds relative to some level of median expenditures on necessities, it is
most consistent if we update in the same way,i.e., with a procedure that changes the thresholds relative
to median expenditures on necessities.
Pros of CPI updating (A or B above):
1. Relative changes create a moving target of poverty. In the short run, one is most interested in
knowing how many people have purchasing power above a pre-set level. If the level changes with
standards ofliving, it's much harder to interpret the poverty statistics.
2. Standard ofliving changes (and social norm changes) occur only very slowly. Better to take them
into account every-so-often, rather than to embed them annually into the poverty rate.
3. Ifwe update with a CPI for necessities only (food, shelter, & clothing), we might capture most of
the relevant changes (since much of the change in expenditures on these items is due to price changes
rather than quantity changes) and have an easier time explaining the short run updating procedure.
Hex-Dump Conversion
Policy Issue 3: Adjusting for Geographic Variations in Cost of Living
Agenda
1. How does the Census propose to do this?
2. What are the results of such an adjustment?
3. What are the alternatives?
A. Adjust for the official poverty counts, but don't adjust the guidelines
B. Publish an adjustment index for researchers who want to use it, but adjust
neither the official poverty counts nor the guidelines
C. Adjust both the official poverty counts and the guidelines
4. What are the criticisms of these different approaches?
**************************************
Reasons to adjust for geographic variation:
If the poverty thresholds provide a measure of "income needs" they should reflect cost of living
differences across regions. Theoretically, most statisticians/economists agree that such adjustments
should be done if the data is available.
Reasons not to adjust for geographic variation:
Adjustments in the thresholds can lead to a "slippery slope", with advocacy groups arguing that such
price adjustments should be done in a wide variety of program benefit payments as well. This can also
lead to rriuch greater efforts by legislators to intervene in exactly how the "correct" regional price
adjustments are done.
Researchers would not be able to duplicate poverty counts based on geographic price variation because
the detailed geographic information required to calculate these would violate privacy constraints in
some cases.
Tentative consensus among the group:
* If Census decides that they want to publish a poverty count based on thresholds with geographical
price variation, we recommend that such an adjustment not be part of the "base" alternative poverty
count, but one of the alternatives.
* At some point in the future, should we discuss changing the poverty guidelines used for program
purposes, these should not include geographical price variation.
Hex-Dump Convers;o~
Policy Issue 4: Accounting for Medical Out-of-Pocket (MOOP) Expenditures
Agenda
1. How does the NAS propose to deal with MOOP expenditures?
Why do it this way?
2. How does this affect poverty calculations?
3. What are the alternatives?
A. Account for MOOP in the resource count (NAS)
B. Account for MOOP in the thresholds
C. Ignore issue entirely
4. How does this issue interact with our measures and data concerning the broader question of health
and health care coverage?
5. What are the pros and cons of these approaches?
**********************************************
Pros of handling MOOP as recommended by the NAS (subtracting it from income):
* The NAS argues (and some of our group strongly agree) that it is better to remove MOOP from the
income side rather than add it into the threshold side.
* This is the approach that Census has investigated and can implement in a timely fashion.
* The alternative of ignoring MOOP entirely (neither taking account of it in the thresholds nor the
income side) has real credibility problems, given extensive attention to the problems oflack of
insurance and high out-of-pocket medical expenditures.
* Ifwe do this adjustment now, further changes that better take account ofMOOP in the future will
involve relatively small changes to the poverty numbers. Ifwe don't do it now and want to take
account of MOOP in the base alternative in the future, it will involve a major realignment of numbers.
Cons of handling MOOP as recommended by the NAS:
* The data for imputing MOOP is not as recent as we'd like (although this will improve markedly in
the next few years.)
* Ifwe ignore MOOP entirely, benchmarking the poverty rate to the 1997 numbers will allow us to set
the thresholds at a share of expenditures that is closer to the NAS recommendations.
* Some of our group are strongly convinced that it would be better to include MOOP in the thresholds
rather *an subtracting it from income, and would like to see this implemented.
This site parses the emails sent and received by Elena Kagan during her time in the Clinton administration and presents them in a more familiar interface.
You may find that some records are garbled or incomplete, or that conversations are fractured. Please bear with us: the source documents are extremely messy. We're working to improve the quality of the documents hosted here.
Due to the programmatic nature of the tools used to build this site, we recommend checking any research effort against the source files released by the Clinton Library.
You are able to view 26,108 of the 29,281 released emails (89.1%)
