Luigi Maria Briglia

Heterogeneous-agent macroeconomics reproduces the cross-sectional and life-cycle distribution of U.S. household wealth but not how households move through it. In PSID data, I identify idiosyncratic expenditure risk: a persistent component of household non-durable consumption (such as medical bills, home and vehicle repairs, or schooling and childcare) that current income, wealth, and demographics leave unpredicted and that does not anticipate higher future income. It is large and persistent, accounting for 15% of the total variance of log consumption with a half-life near nine years. In a calibrated heterogeneous-agent overlapping-generations model with discount-factor heterogeneity, expenditure risk enters as a structural primitive: a persistent shock to the marginal utility of consumption, opening a covariance wedge between the shock and consumption growth in the household's Euler equation. Relative to an income-risk economy calibrated to the same wealth targets, adding expenditure risk closes 78% of the model-data gap on the variance of consumption growth, 85% on escapes from low net worth (non-positive net worth), and 39% on falls into it; the same wedge makes high marginal propensities to consume transient and loosens their tie to current wealth.

How should governments design progressive labor-income taxes when workers can shift labor supply into untaxed informal work? Using household surveys for Brazil, Colombia, Mexico, and Peru, we document steep gradients in informality, employment, and unemployment across the income distribution. We analyze non-linear tax schedules in a heterogeneous-agent model with search frictions, savings, and an endogenous formal-informal labor-supply margin. Progressivity operates through an inclusion margin at the bottom - negative income taxes increase formal attachment - and an evasion margin at the top, where higher marginal tax rates shift labor supply into the untaxed sector. These opposing forces imply that both welfare and formality are hump-shaped in progressivity; in a calibration to Mexico, the welfare-maximizing degree of progressivity is about five times the current level.

We embed extreme-value preference shocks in an otherwise standard consumption-savings problem. Under log utility, the value function and consumption density admit closed forms: the consumption share follows a Beta distribution whose dispersion rises with wealth, matching a robust empirical pattern in PSID micro data. When households can ignore the episodic opportunities generated by these shocks, a selection mechanism emerges: poor households opt out and behave as in the standard model, while wealthy households exercise the opportunities and save more. This wealth-dependent amplification of precautionary saving concentrates the stationary wealth distribution in the right tail without heterogeneous discount factors, stochastic returns, or bequest motives. In a calibrated heterogeneous-agent economy, a single EV-dispersion parameter disciplined by the consumption-error profile accounts for 51-67% of the increase in top wealth shares through the selection channel alone.