What happens to earnings upon marriage? Linking administrative and survey data from Germany, we show that there is a marriage earnings gap. Even after accounting for the child penalty, women’s earnings drop by 20% after marriage. We show that the marriage earnings gap results from both the extensive margin (women stop working) and the intensive margin (women work fewer hours), but not from a decrease in hourly wages. Labor supply disincentives from joint taxation can explain about one quarter of the marriage earnings gap, while we find no effect for labor supply incentives from changes in divorce law. Leveraging variation in norms created by the German separation, we find that gender norms are another important driver behind the marriage earnings gap.

This paper examines monthly earnings volatility and its transmission to household earnings volatility using Norwegian data on the universe of monthly pay histories. We document substantial month-to-month earnings changes: within a job, while over one-quarter of months have no earnings changes, another quarter have at least a 23% change. Accounting for multiple jobs and non-employment increases volatility, while aggregating to households reduces volatility by 12-35%. Event studies around job loss and couple formation, along with decomposition and bounding exercises, show that most of this decline reflects pooling effects rather than sorting or responses to shocks.

A large literature documents that most workers do not opt out from auto-enrollment in pension settings with high savings incentives and complex choices. We show that the effect of defaults is limited in a setting with low incentives, simple options, and easy opt-out. We study the nationwide introduction of auto-enrollment for low-income employees in Germany's public pension system, using administrative data and linked survey data. Auto-enrollment raises participation by {\vsktrddmiallii} percentage points, but the majority of workers actively opt out. The default effect is stronger for workers with higher incentives and for those who lack knowledge of their enrollment status. 

Tax withholding in Pay-As-You-Earn systems is common across developing countries and considered essential for the effective enforcement of personal income taxation. To limit administrative burdens, Pay-As-You-Earn is often coupled with generous tax non-filing options. Drawing on rich tax administrative data for South Africa, we show that such Pay-As-You-Earn systems are associated with sizeable and non-refunded over-withholding of taxes, raising taxpayers' effective tax rates above legally applicable rates. Vulnerable taxpayer groups—low-income workers, young taxpayers and individuals who recently entered the formal labour market—are disproportionately affected. Pay-As-You-Earn over-withholding can deter formal labour supply and impair the progressivity of the personal income tax schedule.

Taxpayers can reduce their tax liability by itemizing deductions and claiming more than the standard deduction they are automatically entitled to. I show that taxpayers also itemize when costs exceed benefits, a behavior that is strictly dominated by not itemizing. Using German administrative income tax data, I document that 60 percent of tax filers make this mistake at least once. They itemize even when the sum does not exceed the standard deduction, or when they already face a zero tax liability before itemizing, or both. The observed itemizing behavior is consistent with a model with optimization frictions in which some taxpayers are misinformed about the benefits of itemizing.