# and Glasscock County in Texas experienced the worst growth of -53% over the twenty

years. Income inequality in 1999 varies much more than inequality in 1979. McPherson

# County in Nebraska had the highest income disparity in 1979 while Holmes County in

Mississippi had the highest income disparity in 1999.

# The average percentage of population living under poverty decreases from 15.8%

in 1979 to 14.1% in 1999 and the maximum value of this variable also decreases from

53% (in 1989) to 52.3% (in 1999). The standard deviation is quite high revealing

considerable variation in the fraction of poor people across counties for both years.

# Tunica County in Mississippi had the highest percentage of poor people in 1979 (52.9%),

and in 1999 it had 33% of poor people. Shannon County of South Dakota had the highest

percentage of poor people living in 1999. All other variables also reveal considerable

variation.

# 5. Regression Results

Table 2 shows the results from the three spatial regressions with the growth rate in

real per capita income between 1979 and 1999 as the dependent variable. Note that in all

the columns I include both inequality and poverty, whose correlation is 0.67. Such a high

correlation between inequality and poverty could lead to multi-collinearity, which

reduces the statistical significance of the variables in the model resulting in lower t-

statistics. However, as Drennen and Saltzman 1998, and Johnston 1984 argue, if the

coefficient estimates of the collinear variables turn out to be statistically significant,

19