Essays on inequality and growth

My first paper surveys the evolution of the Kuznet's hypothesis and recent findings of the inequality and growth literature. Although the negative relation between inequality and growth is now well established in theoretical literature, there are discrepancies in the recent empirical findings. A possible solution is suggested which emphasizes the role of demand patterns created by inequality.; The second paper develops a demand based theory of inequality and growth which accounts for the non-linear relation between them. Inequality affects growth simply through demand patterns for new products when people have hierarchic preferences. I study the link between patent length and growth at different levels of inequality. The incentives to innovate at any time will depend on the demand structure of the economy through potential market size, the length of patents and the endogenously determined growth rate. Since the equilibrium is characterized by a situation where the poor can not afford all the products available on the market, the innovators have a waiting time to enter the market until the poor are rich enough to purchase the new product.; Another interesting outcome concerns redistribution. I find that at high levels of inequality a minimal level of redistribution from rich to poor is needed to increase growth.; In my third paper, I empirically investigate the non-linear relationship between inequality and growth using a new data set and several panel data techniques. Using patent data as a proxy for innovation and two different indices for inequality, I first estimate the theory provided in the previous paper using spline regressions, with fixed effects and dynamic GMM. I then employ several non-linear fits and non-parametric methods. For comparability I also provide the results by using growth of per capita income as a dependent variable. I find a highly non-linear relationship between the level of innovation and inequality. The overall relationship between inequality and growth using the new data set turns out to be negative and significant with all types of estimation methods, although the non-linearity is less obvious.