Results
Table 3 presents the results for all our specifications. All controls have the expected
sign. Results are consistent with conditional convergence; initial per capita GDP
had a negative and significant coefficient on growth. Higher human capital levels
(higher average years of schooling and higher life expectancy at birth) and the initial
price of investment display non-significant parameters, although with the expected
signs. Inequality had a significant negative effect on subsequent long-run economic
growth. Regarding unemployment, none of the two measures considered seem to
have a significant effect on growth. Changes in inequality were not significant. However,
when we let unemployment interact with changes in inequality the interaction term is negatively significant for any of the two variables for unemployment.
Results suggest, therefore, that higher unemployment, when associated to increasing
inequality, has a negative effect on subsequent long-run economic growth 10.
Employment is at the core of recovery and long-run growth. And it is so in particular
because high and persistent unemployment most probably leads to increasing inequalities
that erode growth capacities. On the contrary, when inequality decreases
and this decrease is coupled with large unemployment levels, we understand that this
relationship could be associated to the early stages of development, when urbanisation
is still taking place and high unemployment levels are due to a strong inflow
of workers to the cities. In any case, in our sample we did not find any country with the concurring circumstances
of strong inequality decreases and high unemployment levels. However, we did find
the opposite situation, i.e. increases in inequality linked to increases in unemployment.
Overall, our results point to a strong negative impact of increasing inequality
in association with high levels of unemployment: the third quartile of this interaction