# autoregressive parameter and the spatial error terms are statistically significant, I present

the results only for the SAC model (for both inequality and poverty).

# Column (1) shows that counties with positive growth rate of real per capita

income between 1979 and 1999 experienced low income inequality in 1999. Ten-

percentage point increase in the real growth rate of income is associated with 0.85

percentage point fall in income inequality and the coefficient estimate is statistically

significant at less than the 1% level. Column (1) also shows that counties with a higher

level of income inequality and poverty in 1979 experienced higher income inequality in

1999 and the coefficient estimates are once again statistically significant.

# Column (2) in Table 3 has percentage of people living in poverty in 1999 as the

dependent variable. As with inequality, average growth rate between 1979 and 1999

reduces end of the period poverty and initial levels of inequality and poverty increases

end of the period poverty. The models explain 84% of the variation in income inequality

and 82% of the variation in poverty, respectively.

# While initial levels of income inequality and poverty reduce subsequent economic

growth (Table 2), economic growth reduces end of the period income inequality and

poverty. These two results together indicate that a virtuous cycle of economic growth,

poverty reduction and inequality reduction may exist in the U.S. counties. Thus, for the

# U.S. counties, policies aimed at increasing economic growth need not necessarily face a

trade-off in increasing income inequality.

# Barro (1999) finds that the relationship between inequality and economic growth

depends on whether a country is rich. I test this hypothesis for the U.S. counties by

classifying rich counties as ones having per capita income of more than $17,084 and poor

23