In the paper, we develop a micro-founded model for income inequality that fits the current US evidence and permits discussing the welfare effects of tax reforms given that individuals also adjust their labor supply and human capital accumulation. Heteroscedasticity is crucial in explaining skewed distributions of log-income, while multiplicative risk is necessary for generating Pareto tails. Furthermore, introducing human capital obsolescence as Poisson death jumps also generates Pareto tails in the low end of the distribution and therefore fits the evidence best. According to the model current US taxation is close to its welfare optimum. Major extensions to the model include extending the model from partial to general equilibrium and developing numerical algorithm and code to estimate the extended model.