We study how house price shocks affect marital stability and household labor supply. We address this question using a dynamic collective household model with limited commitment. We find that positive house price shocks increase the divorce rate, and that leverage ratios such as loan-to-income (LTI) and loan-to-value (LTV) determine the transmission of house price shocks on divorce. Given its importance, we then analyze a tightening of the credit market through the LTI-limit. We show that neglecting the divorce and intra-household bargaining channels significantly biases the individual welfare effects of such policies.

We develop an empirical framework to analyze the dynamic effect of personality traits in marriage market patterns and intrahousehold decisions. We exploit detailed information at the individual level from the HILDA survey about consumption, labor supply, time use, and personality traits (as measured by the Big Five). First, we document that personality types are related to marital and divorce patterns, time allocated to both market and non-market labor activities, and the evolution of earn ings. To rationalize these empirical facts, we build a life-cycle model that integrates endogenous household formation and collective household choices under limited commitment. Our framework allows personality to affect both wage processes and individual preferences. In the latter, personality traits enter indirectly through household production and the utility of marriage (match quality). We use the estimates of our model to conduct counterfactuals associated with intrahousehold behavior.

We elicit preferences over ideal childcare allocations and the willingness-to-accept alternatives. Our results demonstrate that the dominant ideal is for young children to spend most of their time at home, with an egalitarian split between parents. Analyzing responses to alternatives reveals that people prioritize children’s time at home with any parent over achieving their ideal parental split, and the aversion to more daycare is substantial: the median willingness-to-accept an alternative with dual full-time working parents and full-time daycare is nearly five times the average cost of daycare in the United States. The ideal level of home care and the aversion to more daycare time is statistically indistinguishable between men and women, suggesting a strong shared preference for parental care, rather than a preference for maternal care.

Nearly half of U.S. marriages end in divorce, yet most spouses plan as though their own will likely not. We field an original survey of married U.S. based individuals and show, against three independent benchmarks, that divorce risk is systematically underestimated: respondents place their own ten-year risk 16 percentage points below that of observably similar peers, 74% fall below a machine-learning benchmark trained on realized PSID marital histories, and two thirds do not revise at all when shown the population divorce rate. We then embed such beliefs in a life-cycle model of household decision-making with endogenous divorce, fertility, and human capital accumulated through learning-by-doing, and discipline the belief process directly with the survey. Optimistic couples specialize more sharply and have children earlier, building less self-insurance against divorce. The resulting gaps outlast the marriage: optimism accounts for 18% of the gender gap in labor income and 28% of the gap in human capital, and raises the poverty rate of women who divorce young by 7.5 percentage points. Joint taxation amplifies the distortion. Beliefs held during marriage thus have an enduring impact on outcomes long after it.

Workers often have substantial discretion in the decision to adopt generative AI. In fact, worker traits are better at predicting adoption than the job. We ask which workers adopt, and why. To do so, we use representative United States survey data to estimate a two-sided matching model with an endogenous technology choice. Workers adopt based on the returns they expect from the technology. This separates adoption into four channels, capturing the actual productive returns, erroneous beliefs about these returns, the enjoyment of using the technology, and learning costs. We find that skills determine adoption through every channel, but mostly through a real productive return. Age and personality run almost entirely through learning costs. Skilled workers understate what they would gain, which dampens the skill gap in adoption. Two policies have opposite effects: better information sends the technology to the workers who gain most but widens every gap in who uses it, while cheaper learning closes those gaps and spreads it to workers who gain less.