Regional Differences in Family Poverty
That poverty rates vary considerably across regions has been confirmed; reasons for these variations need to be understood in order to design appropriate Federal antipoverty policies. This study was undertaken to examine how characteristics of the poor and nonpoor vary among regions and to determine which factors contribute to regional differences in poverty rates.
Who Is in Poverty?
The official U.S. poverty statistics are based on a methodology developed at the Social Security Administration in the 1960’s, which sets income thresholds below which a family is classified as poor. The thresholds vary with total family size, number of family members who are children, and whether the householder is 65 years or older. Each year the thresholds are adjusted for inflation by indexing to the CPI-U. The official U.S. poverty rate in 1994 was 11.6 percent.
The National Academy of Sciences recommends that poverty thresholds vary over geographic units to reflect differences in the cost of housing. Other alternative methods of measuring poverty incorporate additional adjustment factors, such as the value of nonmedical, in-kind government program benefits–food stamps, housing subsidies, and subsidized school lunches. When these factors are considered, the poverty rate in 1994 was 10.4 percent.
Regional family poverty rates are presented in table 1. Patterns of poverty are generally consistent, regardless of measure. The West South Central region, New York, and California are above the national average, whereas New England, the North Central regions, the Middle Atlantic region (excluding New York), the South Atlantic region, the Mountain region, and the Pacific region (excluding California) are consistently below the national average–although their relative rankings vary according to the measure of poverty used.
Table 1. Regional differences in family poverty rates, 1994
Census Poverty rate
geographic adjusted for
division Poverty rate cost of living
Percent
New England 8.2 9.8
Maine, New Hampshire,
Vermont, Massachusetts,
Rhode Island, Connecticut
Middle Atlantic 8.9 9.1
(excluding New York)
New Jersey, Pennsylvania
New York State 14.8 16.1
East North Central 10.7 10.3
Ohio, Indiana, Illinois,
Michigan, Wisconsin
West North Central 9.4 8.1
Minnesota, Iowa, Missouri,
North Dakota, South Dakota
Nebraska, Kansas
South Atlantic 10.6 10.2
Delaware, Maryland,
District of Columbia, Virginia,
West Virginia, North Carolina,
South Carolina, Georgia, Florida
East South Central 13.1 11.0
Kentucky, Tennessee,
Alabama, Mississippi
West South Central 15.8 14.5
Arkansas, Louisiana,
Oklahoma, Texas
Mountain 9.7 9.0
Montana, Idaho, Wyoming,
Colorado, New Mexico, Arizona,
Utah, Nevada
Pacific 8.7 9.0
(excluding California)
Washington, Oregon
California 14.4 17.1
United States 11.6 11.6
Poverty rate
adjusted for
Census cost of living
geographic and in-kind
division benefits
New England 9.4
Maine, New Hampshire,
Vermont, Massachusetts,
Rhode Island, Connecticut
Middle Atlantic 8.1
(excluding New York)
New Jersey, Pennsylvania
New York State 15.3
East North Central 8.9
Ohio, Indiana, Illinois,
Michigan, Wisconsin
West North Central 7.2
Minnesota, Iowa, Missouri,
North Dakota, South Dakota
Nebraska, Kansas
South Atlantic 9.2
Delaware, Maryland,
District of Columbia, Virginia,
West Virginia, North Carolina,
South Carolina, Georgia, Florida
East South Central 9.7
Kentucky, Tennessee,
Alabama, Mississippi
West South Central 12.6
Arkansas, Louisiana,
Oklahoma, Texas
Mountain 8.0
Montana, Idaho, Wyoming,
Colorado, New Mexico, Arizona,
Utah, Nevada
Pacific 8.4
(excluding California)
Washington, Oregon
California 15.9
United States 10.4
Source: Triest, R.K., 1997, Regional differences in family poverty, New England Economic Review, January/February, pp. 3-17.
Regional Differences in Poverty for Specific Groups
Using data from the March 1995 Current Population Survey, this study examines socioeconomic and demographic characteristics of about 40,000 U.S. families and how poverty rates vary over regions for specific subgroups of the population. Table 2 presents poverty rates for families with different characteristics by region.
Table 2. Regional differences in poverty rates for specific groups, 1994
Female head
Census Head with with children
geographic All no high school less than
division families diploma(1) 6 years old
Percent
New England 8.2 19.9 62.9
Middle Atlantic 8.9 21.1 63.2
(excluding New York)
New York State 14.8 36.9 71.5
East North Central 10.7 26.1 64.1
West North Central 9.4 24.5 52.6
South Atlantic 10.6 24.1 58.3
East South Central 13.1 27.1 68.8
West South Central 15.8 36.6 60.1
Mountain 9.7 27.8 63.1
Pacific 8.7 18.5 57.5
(excluding California)
California 14.4 36.5 63.2
United States 11.6 28.7 62.5
Head
Census Black Hispanic immigrated
geographic family family within last
division head head 10 [years.sup.2]
percent
New England 33.4 37.4 21.9
Middle Atlantic 29.2 25.5 17.2
(excluding New York)
New York State 27.9 37.8 41.8
East North Central 32.6 20.0 22.8
West North Central 30.2 21.0 21.6
South Atlantic 22.1 18.8 21.5
East South Central 28.3 20.3 47.0
West South Central 33.8 30.2 41.4
Mountain 28.6 26.3 45.0
Pacific 33.4 20.4 28.2
(excluding California)
California 19.7 28.8 36.1
United States 27.4 27.9 32.5
(1) Computed based only on families where neither the head nor spouse was more than 65 years old.
(2) Individuals were classified as immigrants only if they were born in a country where English is not the dominant language.
Source: Triest, R.K., 1997, Regional differences in family poverty New England Economic Review, January/February, pp. 3-17.
In all 11 geographical areas, the poverty rate for families in which the head does not have a high school diploma was much greater than the overall family poverty rate. The poverty rate for this group was especially high in areas of high overall poverty–the West South Central region, New York, and California.
Families headed by a single woman with at least one child under age 6 had high rates of poverty in all regions; however, the poverty rate was somewhat below the national average in the high-poverty West South Central region and slightly above the national average in low-poverty New England. Highest rates were in New York State and the East South Central region.
Families that have a Black or Hispanic head had high rates of poverty in all regions. For Blacks, highest rates were in the West South Central, Pacific (excluding California), and New England geographic divisions. Lowest poverty rates for families headed by Blacks were in California. Hispanic families had highest rates in New York State and New England, with lowest rates in the South Atlantic region.
Families headed by a recent immigrant from a non-English-speaking country had high poverty rates especially in the East South Central, Mountain, New York State, and the West South Central regions. Poverty rates for this group were lowest in the Middle Atlantic region (excluding New York State).
Regional Differences in the Composition of the Poor
New England and the Pacific regions (excluding California) had relatively low proportions of poor with family heads who lacked a high school diploma (table 3). Families with female heads are a higher proportion of the poor in the relatively low-poverty New England and East North Central regions than in the high-poverty West South Central region and California. Black families make up a higher proportion of the poor (and nonpoor) in the Southern regions than elsewhere. Hispanic families are a higher proportion (53 percent) of the poor (and nonpoor) in California. Immigrant families are a larger share of both the poor and nonpoor in New York and California.
Table 3. Regional differences in the composition of the poor, 1994
Census Head with Female
geographic no high school family
division diploma(1) head
New England 30.8 63.3
Middle Atlantic (excluding New York) 35.4 57.2
New York State 41.4 60.1
East North Central 34.0 62.2
West North Central 32.7 43.7
South Atlantic 39.6 55.3
East South Central 43.1 58.2
West South Central 49.0 43.8
Mountain 36.7 48.3
Pacific (excluding California) 20.2 49.4
California 49.3 41.6
United States 40.3 52.6
Census Black Hispanic
geographic family family
division head head
Percent
New England 17.6 17.2
Middle Atlantic (excluding New York) 29.3 14.8
New York State 30.7 30.4
East North Central 33.1 5.3
West North Central 18.1 2.7
South Atlantic 42.4 9.6
East South Central 41.1 1.3
West South Central 30.0 33.8
Mountain 7.2 38.5
Pacific (excluding California) 9.1 5.9
California 8.6 53.3
United States 27.5 21.4
Head
Census immigrated
geographic within last
division 10 years
Percent
New England 6.4
Middle Atlantic (excluding New York) 4.6
New York State 16.9
East North Central 2.5
West North Central 2.0
South Atlantic 4.1
East South Central .5
West South Central 6.2
Mountain 10.7
Pacific (excluding California) 7.0
California 19.5
United States 7.6
(1) Computed based only on families where neither the head nor spouse was more than 65 years old.
Source: Triest, R.K., 1997, Regional differences in family poverty, New England Economic Review, January/February, pp, 3-17.
Nationwide, most of the poor are either headed by a female (53 percent) or by someone without a high school diploma (40 percent). In New England, New York, and the East North Central regions, over 60 percent of the poor families are headed by a female. Forty-nine percent of the poor families in California and the West South Central region have a head who didn’t finish high school.
These descriptive statistics suggest that both economic and demographic factors help to explain regional differences in poverty rates. Educational attainment varies over regions and is strongly associated with a reduced probability of being poor. Demographic factors such as the family being headed by a single parent or by a member of a minority are associated with an increased probability of being poor.
Relative Importance of Factors Underlying Regional Differences in Family Poverty
A set of probit regressions was used to examine determinants of family poverty status. When only geographic area indicator variables were included, the probability of being poor is lower in New England than in the other areas.
The second regression added variables indicating whether the family is headed by a single woman (and the number of children under age 6 in these families) or by a married couple. A poverty threshold variable was also included in this regression. Since the poverty thresholds increase with family size, a region with a higher-than-average share of large families might be expected to have a higher-than-average poverty rate. However, only small differences in the geographic area coefficients were observed (table 4). Therefore, although these variables are useful in predicting whether a given family will be poor, they are not as helpful in explaining interregional differences in poverty.
Table 4. Family poverty probit regressions(1)
Poverty
Census Division threshold and
geographic effects selected
division only demographic
variables
added
Middle Atlantic (excluding New York) .009 .016
(.010) (.012)
New York State .081 .065
(.012) (.013)
East North Central .032 .027
(.009) (.010)
West North Central .016 .036
(.011) (.012)
South Atlantic .030 .035
(.009) (.010)
East South Central .061 .065
(.011) (.013)
West South Central .091 .105
(.011) (.012)
Mountain .019 .029
(.011) (.013)
Pacific (excluding California) .007 .017
(.012) (.014)
California .075 .072
(.010) (.012)
In (Poverty threshold) .175
(.007)
In (Earnings capacity)
Married couple family .169
(.009)
Female family head .064
(.009)
Number of own children less than 6 .128
in a female-headed family (.005)
Earnings
Census capacity
geographic added
division
Middle Atlantic (excluding New York) .007
(.013)
New York State .059
(.014)
East North Central -.003
(.011)
West North Central .011
(.013)
South Atlantic .009
(.012)
East South Central .018
(.013)
West South Central .053
(.013)
Mountain .000
(.014)
Pacific (excluding California) -.004
(.015)
California .037
(.013)
In (Poverty threshold) .343
(.008)
In (Earnings capacity) -.303
(.005)
Married couple family -.019
(.009)
Female family head .014
(.009)
Number of own children less than 6 .030
in a female-headed family (.006)
(1) For binary variables, the coefficients are changes in the probability of being in poverty associated with the variable being equal to one rather than zero; for continuous variables, the coefficients are the partial derivatives of the probability of being in poverty. Standard errors are in parentheses.
Source: Triest, R.K., 1997, Regional differences in family poverty New England Economic Review, January/February, pp. 3-17.
A third regression added a measure of a family’s potential earnings–the amount that could be earned by adult family members if all of them were to work full time for the entire year. Wage rates were imputed for sample members who were not working. Regional variables were not included in the wage imputations. Thus, the distribution of the earnings capacity measure will vary over regions because of differences across regions in the distribution of workers’ characteristics. How these characteristics are “priced” in the regional labor markets will not affect the distribution of the earnings capacity measure. For example, regional variation in the earnings capacity measure will capture differences in the level of educational attainment across regions but not interregional differences in the economic return to education.
All of the geographic coefficients decrease in size when the earnings capacity variable is included in the specification–most of them, substantially (table 4). Only the one for New York remains statistically significant. Thus, a major reason why other regions have higher poverty rates than does New England is that a larger proportion of their populations have low earnings capacity. Once the ability of families to work their way out of poverty is controlled for, the regional effects are much smaller. Interregional variation in the distribution of human capital appears to be the dominant force in generating the regional disparities in poverty rates.
Further regressions included additional variables that measure constraints in the labor marketplace related to discrimination, poor language skills, or lack of demand (recent immigration, for example). In summary, much of the variation in poverty rates across regions can be explained by variation in the potential earnings of families relative to the poverty thresholds determined by family size and composition. Recent Federal antipoverty policy gives more leeway to States in designing programs, setting limits on the time that individuals can collect benefits, and requiring a greater degree of labor market activity by welfare recipients. Although the primary goal of welfare reform is not to reduce interregional differences in poverty, Government policy can affect differences in the poverty rate across regions. Human capital accumulation may be accelerated with greater involvement of the low-income population in the labor force, but this requires major new training and education programs. Low-skilled workers face increasing competition for jobs in areas that have large welfare populations. In the absence of changes in the interregional distribution of human capital, interregional differences in poverty rates are likely to continue.
Source: Triest, R.K., 1997, Regional differences in family poverty, New England Economic Review, January/February, pp.3-17.
COPYRIGHT 1998 Superintendent Of Documents
COPYRIGHT 2004 Gale Group
