increase in initial income inequality is associated with approximately thirteen-percentage
point decrease in economic growth. The coefficient estimate is statistically significant at
less than the 1% level. Studies on Unites States county-level growth also find a similar
result (Rupasingha et al. 2002).
The coefficient estimate on initial level of poverty is also negative suggesting that
counties with a higher initial level of poverty experienced slower economic growth
between 1979 and 1999. Counties with ten-percentage point higher level of poverty in
1979 experienced 0.07 percentage points decrease in economic growth, compared to
others, ceteris paribus. Once again, the coefficient estimate is statistically significant at
less than the 1% level. This result is consistent with Bhatta (2001) who finds that initial
level of poverty negatively affects economic growth in the U.S. Metropolitan Statistical
Areas. Thus, both initial levels of income inequality and poverty reduce economic
growth. Growth rate of human capital and population (between 1979 and 1999),
percentage of white population, percentage of people employed in manufacturing sector
and per capita government expenditure in 1979 also positively affect economic growth.
All the above coefficient estimates are statistically significant as well.
The coefficient estimate of the autoregressive parameter suggests that ten-
percentage point increase in growth in a county increases economic growth by
approximately 2.1 percentage point in its’ neighboring county. Rupasingha et al. (2007)
get almost the same value for the autoregressive parameter, which is also significant at
the 1% level.11 This outcome suggests the presence of spill over effects of economic
growth in neighboring counties. The highly significant spatial error coefficient implies
The value they get is 0.205 and mine is 0.21