View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

November 15,1994

Federal Reserve Bank of Cleveland

Understanding Differences
in Regional Poverty Rates
by Elizabeth T. Powers and Max Dupuy

Ahe U.S. poverty rate is among the
most widely used indicators of the
nation's economic success—and of how
that success is shared. However, the
national rate masks tremendous variation
in the regional numbers. In 1992, for
example, the official poverty rate ranged
from a low of 8.3 percent in New England to a high of almost 16 percent in the
East South Central states.
If we accept the poverty rate as a legitimate yardstick of national achievement,
then policymakers have several reasons
to be concerned about this disparity.
First, interregional equity is a legitimate
policy concern in and of itself, analogous
to concerns about how income is distributed among individuals. In fact, interregional poverty differentials may be
another dimension along which the wellbeing of the nation is judged—a premise
that casts doubt on the desirability of
having means-tested cash transfers (commonly termed "welfare") determined by
the states. Second, even if equalization
of poverty across regions is not a policy
goal, understanding the marked and persistent differences in the regional statistics
can shed light on the factors contributing
to poverty. And finally, it is important to
determine whether interregional disparities truly reflect meaningful differences
in well-being.

Poverty is the product of a multitude of
factors, including market conditions,
demographic characteristics, and fiscal
policy. Its accurate measurement is complicated by interregional differences in
the cost of living and the quality of life,
among other factors.
A common feature of all these determinants is their great variation across
regions. The burdens of recession are
generally distributed unevenly over the
country, as illustrated by the disparate
experiences of the West and Midwest
during the last downturn. Demographic
types that are overrepresented among the
poor, such as female-headed households,
are more prominent in some areas than
in others. State fiscal policies, which are
presumably uncoordinated, are an obvious source of interregional poverty variation. Federal fiscal policy also plays a
role through both intended and unintended interregional transfers. Finally,
the official poverty level is not adjusted
for regional cost-of-living differences,
which may contribute to large and persistent disparities in measured rates.
Recent legislative interest has made it
particularly important to understand how
such adjustments might change perceptions of the poverty picture, since the
flow of federal dollars to the states would
be affected by changes in the measured
distribution of poverty.

The official U.S. poverty rate, often
cited as a measure of our nation's
economic strength, hides a huge variation across regions. Here, the authors
examine the reasons for this disparity,
focusing on differences in demographic,
economic, policy, and cost-of-living
factors across the nine U.S. census
divisions. Results show that although
all of these elements have some influence on the poverty gap, economic
factors are key.

In this Economic Commentary, we
employ a simple method (see box) to
attribute poverty differences across the
nine standard census regions to demographic, economic, policy, and cost-ofliving factors.1 Our results show that
economic factors are the primary source
of the disparity, with weak economies
accounting for a large share of the aboveaverage poverty rates seen across the
United States. Surprisingly, government
fiscal policy, although poverty-reducing
on an absolute basis in every region, is
less effective in areas with weak economies. Our findings also show that
demographic patterns play an important
role. Interestingly, a crude cost-of-living
correction would result in a very different
poverty picture, implying a major change
in the flow of funds from the federal
government to state coffers.

Our method for decomposing differences in regional poverty rates can best be illustrated by working through a comparison of the East North Central (ENC) region and
the United States. First, families are sorted into three classifications: female-headed
households with children, two-parent families with children, and other family units.
Data from the March Current Population Survey are used to compute the fraction of
each type of household in poverty in both the ENC states and the United States as a
whole. To compare the difference in poverty rates attributable to demographic characteristics peculiar to the ENC, we compute the total poverty rate using U.S. familyspecific rates, but apply the ENC proportions of the three types. Comparing this rate
with the actual U.S. poverty rate reveals whether poverty in the ENC is above or below

account for a significant portion of the
total value of transfers, particularly for
low-income families.4 Consequently, we
count as income the value of some normally excluded in-kind benefits.5



New England

8.30 (9)
10.80 (5)

10.29 (5)


East North Central
West North Central

10.27 (7)

9.68 (7)
9.09 (8)

7.92 (9)

South Atlantic
East South Central
West South Central


14.20 (2)

10.08 (5)

This more complete measure does not
change the basic picture of interregional
poverty. Table 1 presents the official
family poverty rate for the nine census
regions in column 1 and our measure in
column 2. Official rates range from a
low of 8.3 percent of families in New
England to more than 15 percent in the
South Central divisions. As has historically been the case, the South has the
highest poverty rates by far. The Northeast has the lowest overall rate, thanks to
the influence of New England. Poverty
rates in the West have been rising
steadily and now surpass those of the
Midwest. The difference between the
rates in columns 1 and 2 is due entirely
to tax and in-kind transfer policy. Note
that all of the adjusted rates are lower,
indicating that the effect of in-kind transfers dominates, but that the poverty
ranking of the divisions remains
unchanged. Unless otherwise stated, the
term "poverty rate" will refer to this taxand transfer-adjusted measure.


10.36 (6)

9.85 (6)

9.31 (7)

Decomposing Poverty
Differences: An Overview

United States




The first step toward understanding
regional poverty differentials is to
decompose them into their demographic,
policy, and economic components. We
compare regional poverty to the national
rate for income reported in 1992, the
latest year for which data are available.6
One advantage of using 1992 data is that
it was a fairly typical year—neither a
peak nor trough of the business cycle (at
least not nationally), and free from any
major fiscal policy changes that would
dramatically affect the level or distribution of poverty rates. Table 2 compares
each division's family poverty rate to the
national average and, implicitly, to each
other, based on regional characteristics.
(The divisions are listed from least to
greatest poverty for convenience.) The
first column is the percent deviation of
the region's rate from the nation's. A
negative number indicates that the
regional rate lies below the national

average due to the special demographic characteristics of the region. Dividing this
difference by the U.S. poverty rate yields the entry of 3.25 percent for the ENC
demographic factor in table 2. Policy effects are derived similarly.
After Taxes, Transfers,
and Some In-Kind Benefits

Adjusted for
Cost of Living

NOTE: Regional rankings are in parentheses.
SOURCE: Authors' calculations based on data from the March 1993 Current Population Survey.


Measuring Poverty

In the late 1950s, the federal government
developed poverty thresholds based on
cash income. At that time, data revealed
that the average American family devoted
about one-third of its budget to food.
Thus, a subsistence food budget was
multiplied by three to yield a subsistence
income, or poverty line. The poverty line
varies by family size, composition, and
ageof householder. Larger families are
assigned higher poverty thresholds, while
older families face lower ones. With
minor modifications, these guidelines
have been in use since 1961 and are
updated yearly using the Commerce
Department's Consumer Price Index for
all items. The official family poverty

rate is determined by computing the
fraction of all families whose pre-tax
cash income falls below their familyappropriate levels.2
Since the development of the original
poverty statistic, there have been two
important policy changes that might
affect the measurement of poverty. First,
the tax burden on low-income families
has increased since the late 1950s, when
it was about zero.3 Thus, we calculate our
poverty measure based on after-tax and
cash transfer income. Second, "in-kind"
transfer programs (primarily Medicare,
Medicaid, food stamps, and housing
subsidies) that did not exist in 1961 now


New England
West North Central
East North Central
South Atlantic
West South Central
East South Central

BY FACTORS, 1992 a

» -10.21




a. National average = 10.97.
b. Residual after demographic factors and government taxes and transfers are taken into account.
SOURCE: Authors' calculations based on data from the March 1993 Current Population Survey.

average, while a positive number indicates the opposite. For example, New
England has a family poverty rate of 7.7
percent, about 30 percent below the
national average of 10.97.
These divisional deviations from the
U.S. average can be attributed to several
factors. The demographic factor conveys
the role of the relative regional distributions of family types in poverty. For
example, the West North Central division's poverty rate is 4.5 percent below
the national average due to the region's
more favorable demographic conditions
(primarily an above-average concentration of two-parent families), accounting
for about one-quarter of the total difference. The "government policy" factor
measures the net impact of all federal,
state, and local taxes and transfers on
poverty. Government policy appears to
do a better-than-averagejob of poverty
relief in New England, for example,
reducing its rate nearly 10 percent below
the national level and accounting for a
full one-third of the difference between
the region's rate and the U.S. average.
Finally, "economic conditions" captures
both transitory (business-cycle-related)
and long-run regional differences in economic status. The effects are not surprising: Relatively strong economies have
below-average poverty, while less-thanaverage economic performance boosts
the poverty rate in the two poorest
southern divisions from 22 to nearly 30
percent above the national average.

To get an overview of the primary
causes of regional poverty differences,
it is convenient to group the nine census
divisions into those with below-average,
average, and above-average poverty. We
designate the four regions with poverty
rates less than 90 percent of the national
average as the below-average group—
New England and the West North Central, East North Central, and Mountain
states. Demographic compositions
favorable to lower poverty are fairly
important in New England and the West
North Central area. Unfavorable demographics are overcome by other factors
in the East North Central division, and
have little effect in the Mountain states.

Finally, the West and East South Central
poverty rates are 129 percent and 136
percent of the national rate, respectively.
Demographics can account for only a
modest portion of the disparity in the
East South Central division, which has
the highest concentration of single-parent
families in the nation. Government policy also has somewhat below-average
effectiveness at reducing poverty in both
divisions. The most significant influence
on the huge poverty gap between the
South Central states and the rest of the
country appears to be a less robust economy: More than 75 percent of the area's
deviation from the national rate is
accounted for by economic conditions.

What all four divisions do have in
common is above-average economic
performance. Perhaps surprisingly, the
contribution of government policy to
below-average poverty in New England
and the North Central states is nearly as
large as that of relative economic success.
The salutary effects of government policy
are particularly important in the East
North Central division: If policy had its
typical impact, poverty rates there would
be little better than the national average.

• A Closer Look at
Government Policy

The poverty status of the Mid-Atlantic,
Pacific, and South Atlantic divisions
follows the national average quite closely.
While these areas appear to be similar to
the nation in demographic, policy, and
economic factors, we shall see that the
Pacific region is actually much different
from the national norm along several

While we have seen that government
policy plays an important role in areas of
extreme poverty, its overall impact can
mask crucial differences between state
and federal policy. The detailed components of the "government policy" category from table 2 are presented in table
3. Both federal and state policies exert
large and often offsetting influences on
poverty rates.7 However, because federal
and state taxes do not have great differential impacts on poverty over regions,
we do not describe their effects in detail.
States have primary responsibility for
cash welfare payments, so state transfer
policy is an obvious source of interregional poverty differences.8 Consistent
with the known facts about welfare,


New England
West North Central
East North Central
South Atlantic
West South Central
East South Central

Total State
Total Federal
and Local
State and
State and
Government Government
Government Local Transfers Local Taxes








a. National averages 10.97.
SOURCE: Authors' calculations based on data from the March 1993 Current Population Survey.

table 3 illustrates that the poorest states
also have relatively less effective transfer
programs. Part of this can be explained
by the wider gap between pre-transfer
income and the poverty line in these
regions.9 That is, poorer regions have
relatively more individuals whose pretransfer income is well below the poverty
line, whereas relatively better-off regions
have more people whose pre-transfer
income is near the line. Hence, even if
all nine divisions offered identical welfare benefits, the poverty-reducing
effect of state transfer programs would
be smaller in the South. This effect is
reinforced by state policy, since southern
welfare benefits are known to be well
below the national norm. Evidence from
table 3 also indicates major differences
in state transfer policy, as can be seen
from the wide variation in the effectiveness of state transfers for areas with
similar poverty rates (for«xample, the
Pacific and South Atlantic divisions).

federal welfare program—food stamps
—helps to equalize the generosity of
state welfare programs, thus reducing
interregional inequality, the vast majority of federal transfers are not welfare
but rather non-means-tested cash benefits
(primarily Social Security) that remove
many elderly from the nation's poverty
rolls. Since Social Security benefits are
based on lifetime work experience and
average wages, higher benefits flow both
to areas with solid labor market histories
(as table 3 makes apparent for the two
midwestem divisions, where poverty
is reduced more than 5 percent below
the national average by federal transfer
policy) and to retirement havens, particularly in the South Atlantic and Mountain
regions. The Earned Income Tax Credit
has become a significant transfer program
to the poor and near-poor, but because it
targets workers, it does not have a neutral
poverty-relieving effect across regions

Table 3 also shows that the effectiveness
of federal transfer policy varies a great
deal across census regions. Federal
transfer payments to individuals are
made for the most part without regard to
residence. The question, then, is what
accounts for this tremendous variation?

The Pacific region is a case where large
policy effects are masked at the aggregate level. Federal transfers have the
weakest poverty-reducing effect there,
but this is almost completely offset by
state and local transfers. However, the
outcome of these opposing forces is not
a neutral government policy. The Pacific
states have a well-below-average poverty
rate for single-parent families (due to
relatively more generous state transfer
policies), but the plight of the elderly
and others counting on federal dollars
appears to be relatively worse.

As before, there is the measurement
issue based on the poverty "gap." However, it is also clear from comparing
regions with similar poverty rates that a
wide disparity exists in the effectiveness
of federal transfers. There are several
reasons for this. Although the primary

• Adjusting for
Cost-of-Living Differences
Poverty guidelines have long been criticized for ignoring regional differences
in the cost of living. If, for a given
amount of income, a family can attain
more or better food and clothing, a larger
apartment, etc., in Iowa than California,
then perhaps having a single poverty line
for both states makes little sense. Based
on recent cost-of-living estimates, the
California poverty line would presumably have to be set as much as 20 percent
higher to enable the state's lowestincome families to consume the same
quantity and quality of items as does an
officially poor family in Iowa.
Recently, policymakers have expressed
interest in implementing state-specific
poverty lines. Because federal aid for
the Head Start program, community
development block grants, home energy
assistance, and remedial education for
poor children is distributed to states
according to their poverty rates, the
potentially large changes to relative
rates from a cost-of-living adjustment
are of more than just academic interest.
To explore the poverty implications of
regional price indexing, we use an index
of state prices to adjust the official poverty line in each state.10 The resulting
rates are presented in the third column of
table 1. Because the poverty line is lowered in states with below-average costs


West North Central
New England
East North Central
South Atlantic
West South Central
East South Central




Adjusted for
Cost of Living







a. National averages 10.97.
b. Residual after demographic factors and government taxes and transfers are taken into account.
SOURCE: Authors' calculations based on data from the March 1993 Current Population Survey.

and raised in states with above-average
costs, poverty rates decline in the Midwest and South and rise dramatically
for the high-cost coastal areas. These
changes are large enough to alter the
poverty status of some divisions significantly. The Pacific region, which contains the high-cost states of California,
Alaska, and Hawaii, zooms from a position of average poverty to the poorest of
all divisions. By contrast, the low-cost
West South Central division, formerly
among the poorest areas, boasts a poverty rate little different from the national
average after cost adjustment. The New
England and East South Central divisions are so firmly ensconced at the
poles of poverty that they remain near
the extremes even after huge cost-ofliving adjustments.
Table 4 recreates the calculations from
table 2, but includes the cost of living as
an additional factor. Because poorer
areas tend to have lower costs and richer
areas higher costs, overall interregional
inequality is reduced. With the exception of the Mountain and West North
Central states, areas with below-average
costs also demonstrate below-average
economic performance. Thus, extremely
high poverty rates no longer appear to
be a "southern problem," because
although economic forces continue to
drag down the South Central regions,
cost-of-living adjustments substantially
mitigate these forces.

This different picture of poverty should
be interpreted cautiously, however. Living costs also reflect the presence of
amenities that people value, such as
warm weather and proximity to recreational resources. As is true for any other
desirable goods, high demand bids up the
price of access to these items. Although
it is difficult to put a price tag on amenities, they make some contribution to
individuals' well-being—including the
low-income population—implying that
it is probably inappropriate to adjust the
poverty lines of any two regions by the
full, conventional cost-of-living differential. The true interregional distribution
of poverty may be intermediate between
the extremes of columns two and three
in table 1.
It is tempting to conclude that an appropriate state-by-state cost-of-living
adjustment would result in a superior
measure of poverty. However, this is not
necessarily the case, since the cost of
living within a state often varies by more
than that between states. This is particularly true when comparing rural and
urban areas. For example, rural family
poverty in a high-cost state might be
overstated, while urban poverty in an
overall low-cost state might be understated. These biases could be as great or
greater than those that arise from applying the same poverty line cutoff to the
entire country. The point is not that these
adjustments could not be made, but that
crudely deflated poverty measures, such
as those presented here and elsewhere,
should be viewed as highly preliminary

and are probably not an appropriate
basis for policy.



Cost-of-living adjustments aside, disparities in economic conditions across
states are the most important determinant of regional poverty variation in the
United States. Some of these economic
differences reflect transitory economic
shocks whose effect on income might
appropriately be smoothed by government policy. More persistent differences
in economic performance—such as the
long-standing gap between the South and
other areas—are well-known puzzles.
Whether these permanent differences
should engender permanent redistributory schemes depends on one's beliefs
about why these gaps exist and whether
alternative policies would provide better
Another significant finding is that current government transfer policies tend
to have below-average effectiveness in
poorer regions, enhancing measured differences in regional poverty rates. Social
Security, the most effective povertyreduction program by far, obviously is
not geared toward interregional income
smoothing, but rather is linked to recipients' past labor market performance. It
seems unlikely that much sentiment
could be aroused among policymakers
or the public for reducing the unequal
regional impact of federal transfers on
measured poverty through changes in

non-welfare programs. One alternative
would be to federalize the welfare system, since the poverty-reducing effectiveness of state transfer policy varies
greatly from region to region. This
would not be a complete solution, however, because most of the differences
seen in regional poverty rates are not
attributable to state welfare programs.



1. The census divisions are as follows: New
England (Maine, New Hampshire, Vermont,
Massachusetts, Rhode Island, Connecticut),
Mid-Atlantic (New York, New Jersey, Pennsylvania), East North Central (Ohio, Indiana,
Illinois, Michigan, Wisconsin), West North
Central (Minnesota, Iowa, Missouri, North
Dakota, South Dakota, Nebraska, Kansas),
South Atlantic (Delaware, Maryland, Washington, D.C., Virginia, West Virginia, North
Carolina, South Carolina, Georgia, Florida),
East South Central (Kentucky, Tennessee,
Alabama, Mississippi), West South Central
(Arkansas, Louisiana, Oklahoma, Texas),
Mountain (Montana, Idaho, Wyoming, Colorado, New Mexico, Arizona, Utah, Nevada),

Federal Reserve Bank of Cleveland
Research Department
P.O. Box 6387
Cleveland, OH 44101
Address Correction Requested:
Please send corrected mailing label to
the above address.
Material may be reprinted provided that
the source is credited. Please send copies
of reprinted materials to the editor.

and Pacific (Washington, Oregon, California,
Alaska, Hawaii).
2. For example, in 1992, the poverty threshold for a two-parent nonelderly family with
two children was $14,228.
3. After peaking in the mid-1980s, the tax
effect on poverty has declined, however.
4. These programs deliver benefits in the
form of vouchers for goods and services
rather than cash.
5. We exclude Medicaid and Medicare, since
the determination of their equivalent values
in cash remains controversial.
6. This article was written before revised
population weights for the March 1993 Current Population Survey became available.
7. Local transfers and taxes are included in
the state categories.
8. Unemployment insurance and workers'
compensation payments are also included in
state transfers, but because they have a minimal impact on poverty rates, we do not discuss them separately.

brought up to the poverty line, which reduces
the effectiveness of transfers in areas with
low pre-transfer poverty.
10. None of the available price data are ideal
for such an adjustment. We use an index constructed by the American Federation of
Teachers from data collected by the American
Chamber of Commerce Researchers Association. Since costs are assumed to be those of a
middle-manager's family, the index is somewhat inaccurate for poor families.

Elizabeth T. Powers is an economist and Max
Dupuy is a research assistant at the Federal
Reserve Bank of Cleveland.
The views staled herein are those of the
authors and not necessarily those of the Federal Reserve Bank of Cleveland or of the
Board of Governors of the Federal Resen'e

9. This bias is somewhat mitigated, however, since areas of low poverty have relatively fewer people who can potentially be

U.S. Postage Paid
Cleveland, OH
Permit No. 385