The following report, funded by the Appalachian Regional Commission (ARC), explores recent poverty trends for the 399 counties that comprise Appalachia, and examines the Census Bureau's Small Area Income and Poverty Estimates' [i] effects on the ARC distressed county designation. We begin with an examination of the changes in total poverty in Appalachia between 1979 and the mid-1990s, with particular emphasis paid to the post-1990 period. The gap in poverty between Appalachia and the rest of the country declined as poverty outside Appalachia increased during the 1980s while remaining virtually unchanged in Appalachia. The U.S. average poverty rate declined from 15.1 percent to 13.8 percent between 1993 and 1995, while poverty among Appalachian counties declined from an average of 16.1 percent in 1993 to 14.6 percent in 1995. The ARC counties with relatively higher rates of poverty are generally concentrated in Kentucky, as well as West Virginia, southern Ohio, and Mississippi. Although Appalachia has long been struggling economically, Appalachia's total poverty rate in 1995 was only slightly higher than in the rest of the country.

Child poverty in Appalachia increased slightly between 1989 and 1995, following the national pattern. In particular, young children in Appalachia have experienced the greatest increases in poverty, compared with older children and the general population. The geographical patterns of total poverty and child poverty are overwhelmingly similar, with higher rates of child poverty concentrated in eastern Kentucky, and significant portions of northern Tennessee, West Virginia, southern Ohio, and Mississippi. Between 1993 and 1995 relative increases in child poverty were most expansive in Alabama, the Carolinas, and New York, followed by Kentucky, West Virginia, Virginia, Pennsylvania, Mississippi, and Georgia. Only Ohio and Tennessee experienced fairly consistent relative declines in child poverty during the period. Similar to the overall poverty rates for the sub-regions, the Central sub-region continued to experience the highest child poverty rates within Appalachia. More than one-third of the children who lived in the Central sub-region lived in households with incomes under the poverty line, with the poverty rates in the other regions just over 20 percent. Poverty rates for children ages 0-4 years were, and continue to be, considerably higher than for children ages 5-17 years both nationally and in Appalachia. This gap was even wider for Appalachian counties than for the remainder of the U.S., with 27.3 percent of children ages 0-4 in poverty, compared to 19.5 percent for children ages 5-17 in 1995.

The ARC has used the distressed county designation for almost twenty years to identify counties with the most structurally disadvantaged economies. Each year the ARC updates the distressed status of counties based on more current information on unemployment and per capita market income. However, reliable county-level poverty rates have, until recently, only been available from the decennial census at the beginning of each decade. The Census Bureau SAIPE program has produced county-level poverty estimates for 1989, 1993 and 1995, giving the ARC the option of using more recent poverty data to classify counties. We evaluate the influence of post-censal estimates of poverty on the traditional distressed county classification, which uses only the estimates of poverty from the most recent census, during both the 1980s and the early 1990s. Of the 399 Appalachian counties, the number designated as distressed increased between 1980 and 1990 by 50 percent. [ii] This increase reversed a two-decade decline in the number of distressed counties. Changing relative poverty levels were a factor in 10 of the 12 transitions out of distressed status during the 1980s. Poverty did not contribute quite as greatly to the much larger number of counties that became distressed in the 1980s.

Principally, we use two analyses to evaluate the viability of the SAIPE for the ARC designation of distressed counties. We first evaluate the accuracy of the distressed status designation at the end of a decade, comparing the 1980 census with the 1989 SAIPE (using the 1990 census as the standard of accuracy). With certain caveats, the results from the 1980s demonstrate that as a decade progresses, the SAIP point estimates more accurately predict the status of both distressed and non-distressed counties than the poverty estimates from the previous census. Then we examine the causes of status transitions that would occur in the early 1990s incorporating the SAIPE into the distressed county designation. The number of counties that have been affected by economic change in the 1990s can be better evaluated and joint changes in unemployment, income, and/or poverty can be distinguished from changes in poverty alone. Between 1990 and 1994 the number of distressed counties in Appalachia declined sharply (38 percent), due more to overall economic improvement in Appalachia relative to the U.S. as a whole than by substitution of the SAIPE for the 1990 census poverty estimates. Moreover, relative shifts in unemployment played a more important role as an independent cause of these transitions out of distressed status than did shifts in poverty.

The distressed status accuracy results from the end of the 1980s suggest that the SAIPE would provide a better determinant of distressed status than the poverty estimates derived from a decade old census. The magnitude and causes of distressed status transitions in the first half of the 1990s indicate that using the SAIP estimates would alter the counties that would be designated distressed by the ARC but not to a radical degree. However, both of these analyses demonstrate that a simple substitution of the SAIP point estimates for census poverty estimates may unjustifiably deny some counties distressed status recognition. As an antidote to this situation it might be more defensible to combine the SAIP point estimate and the SAIP upper bound estimate in the future determination of distressed status. This would accomplish the objective of utilizing more current estimates of poverty while reducing the negative consequences of utilizing an estimate of poverty with greater statistical variation than decennial census derived estimates.

Overall, the analysis of the 1990s indicates that the number of distressed counties has declined in Appalachia during the decade. The Small Area Income and Poverty Estimates indicate a decline in poverty in Appalachia relative to the U.S. as a whole, which reflects a concomitant relative decline in unemployment and a relative increase in per capita market income. Determination of distressed status using the 2000 Census of Population and Housing poverty rates should confirm this decline. During the next decade, the accuracy of the SAIPE program should improve significantly as new sources of income and poverty data, especially the American Community Survey (ACS), become available, making them an even more viable option for the determination of distressed status by the Appalachian Regional Commission.

Notes

[i] The Census Bureau's Small Area Income and Poverty Estimates are abbreviated as SAIPE. These will also be referred to as "SAIP estimates" to focus on the numerical estimates themselves rather than the overall statistical estimates program

[ii] The number of distressed counties in 1990 does not correspond to the number of counties officially designated distressed by ARC because distress levels were frozen during the 1988-1992 period awaiting the release of 1990 census poverty data (Wood and Bischak 2000). The distressed designation uses three year averages of unemployment and per capita market income. Numbers in Table 4.1a are based on a formula for defining distressed counties that incorporates poverty estimates from the last census, not the Census Bureau's post-censal SAIPE estimates.