How do portfolio choices and asset prices impact inequality when agents have stock market participation constraints? To answer this question, we develop a new methodology for using deep learning to characterize global solutions to macroeconomic models with long-term assets, agent heterogeneity, and household optimization under financial frictions. We first characterize the equilibrium recursively in the space of wealth shares and then show how to train neural networks to approximate the equilibrium. This approach extends deep learning tools to a general class of macro-finance models. We use our toolkit to study how asset market participation constraints impact inequality. Market segmentation generates endogenous volatility, which allows wealthy agents to take greater of advantage high expected returns during a recession and amplifies inequality.

Contrary to the common assertion that households have little impact on stock prices, we find their relevance is of first order. We quantify their impact using an asset demand system applied to the complete ownership data for all Norwegian stocks from 2007 to 2020. Households contribute the most to stock market volatility relative to their market share. Even in absolute terms, they come second, surpassed only by institutional investors. Our granular data on households reveal a strong factor structure in household demand: The demand of the rich is distinct from less affluent investors, accounts for the bulk of volatility attributable to households, tilts away from ESG, and is informative about future firm fundamentals. We conclude by using the demand system to measure the profits one can make from trading on household demand shocks.

We investigate the interplay between asset prices, wealth inequality, and taxation in a dynamic general equilibrium economy populated by multiple agents with heterogeneous risk aversions. Tax revenues are collected from consumption taxes and are equally redistributed to all investors through non-pledgeable government transfers. Taxes address wealth inequality by ensuring stationarity of consumption share distributions and preventing consumption shares of less affuent investors from diminishing toward zero. Higher taxes increase stock risk premia and volatilities by shifting wealth toward poorer risk-averse investors, and tend to decrease stock price-dividend ratios and interest rates. The rise in risk premia and decrease in interest rates benefit more affluent, less risk-averse investors, partially offsetting the impact of higher taxes on their wealth, albeit to a small extent. We find that taxes do not prevent high concentrations of wealth at the top of the wealth distribution due to the investment decisions and tax responses of more affluent investors. We also extend the model to incorporate restricted stock market participation and show that its interplay with taxation increases stock risk premia and volatility, and decreases interest rates.