Skills learned on the job vary in their portability across employers. Some are portable and retain their value when workers change jobs, while others are firm-specific and are only productive at the employer where they were acquired. This paper studies how differences across firms in the portability of skills learned on the job shape wage inequality within occupations and wage losses following displacement. I develop an equilibrium search model in which firms differ in the extent to which they equip workers with portable versus firm-specific human capital. I estimate the model using French administrative data that link workers' careers to firm balance sheets and closures. The estimated share of on-the-job learning that is firm-specific averages 55%, with substantial heterogeneity across firms. Counterfactual simulations show that equalizing this share across firms at its baseline mean reduces within-occupation variance in log wages by 18%. This decline is concentrated among workers who enter the labor market with fewer skills and whose wages depend more on human capital accumulated on the job. Eliminating cross-firm differences in the portability of learning also reduces average wage losses two to four years after displacement by 23%.

We study how the specification of wage risk in quantitative life-cycle models changes the welfare evaluation of taxes and transfers. We estimate four wage processes: a canonical linear process, a flexible nonlinear process with age dependence, non-normality, and state-dependent persistence, and two intermediate processes; comparisons across the four isolate the role of each feature: age dependence, non-normality, and nonlinear persistence. The estimated processes imply different allocations of wage risk across permanent, persistent, and transitory components: the richer processes assign less risk to short-lived transitory shocks and more to long-lived permanent and persistent components. We embed each process in the same incomplete-markets life-cycle economy with endogenous labor supply. Removing federal tax progressivity and means-tested transfers raises mean welfare by 0.72% of lifetime consumption under the nonlinear process but lowers it by 0.47% under the non-normal process. These welfare differences reflect both the total amount of wage risk and its allocation across permanent heterogeneity, persistent shocks, and transitory shocks. Holding this amount and allocation fixed, non-normal shocks create rare but severe low-income states that increase the value of public insurance, while the state-dependent mean reversion that characterizes the nonlinear specification works in the opposite direction: unusually bad persistent states tend to be undone by subsequent favorable shocks, so the wage process partly insures itself, and that lowers the value of public insurance.

How does globalization affect workers' wealth and welfare? To answer this question, this paper develops a dynamic open economy model with search frictions and endogenous savings. Heterogeneous, risk-averse workers carry wealth to smooth their consumption during unemployment spells, but borrowing constraints leave low-wealth households with uninsurable income risk. These risks are heightened by globalization, which increases income volatility and reduces job security. We fit the model to data from Colombia, and use it to quantify the welfare gains and losses from globalization for different types of households. Counterfactual experiments show that lowering trade barriers results in an average welfare gain of over 2%. However, while workers who begin their working lives with wealth below the median incur welfare losses of around 2%, those who start at the top of the wealth distribution see significant increases.

House prices shape household consumption, particularly when housing is illiquid and households are highly indebted. This paper examines whether borrower-based macroprudential policies can mitigate credit-fueled housing booms and busts and improve welfare. I build an equilibrium macro-housing model with heterogeneous households, a frictional housing market, and long-term mortgages, and calibrate it to Spain's housing boom and bust of 1998–2014. The model shows that caps on loan-to-value, debt-to-income and payment-to-income ratios substantially dampen the bust in house prices and consumption, raising welfare at the cost of a smaller boom. These gains arise partly because lower leverage limits defaults and forced sales, mitigating the price collapse and its spillovers to other households. The instruments are not interchangeable: a debt-to-income cap delivers the largest welfare gain in the Spanish baseline, while combining instruments performs best across alternative boom scenarios. The results show that the effectiveness of borrower-based policies depends on how their borrowing limits respond to the forces driving the housing cycle.

Macroprudential authorities regulate mortgage risk on two sides: borrower-based limits restrict how much households can borrow, while capital requirements determine how well banks can absorb losses when mortgages default. This paper studies how the two interact. We build an equilibrium model in which heterogeneous households borrow, sell, and default in a frictional housing market whose liquidity evolves over the cycle, and banks fund mortgages with deposits and equity subject to capital requirements. When homes become harder to sell, defaults rise and erode bank capital, and weaker bank capital in turn constrains lending. We use the model to study three channels through which the instruments may complement each other: lower default risk reduces the capital banks need, releasable buffers can sustain lending in a bust, and the instruments differ in who bears their cost, as capital requirements raise borrowing costs for all households while borrower-based limits fall only on the most leveraged buyers. The framework allows us to identify which combination of borrower- and lender-based measures best stabilizes housing cycles and improves welfare, and when using both does better than relying on either alone.

Workers in stable jobs are changing employers less often than they did in the 1990s and 2000s. At the same time, firms are capturing a growing share of the productivity gains workers generate over the course of their employment relationship. Both trends are particularly pronounced in industries that have adopted new technologies more intensively. This paper argues that these trends reflect a common mechanism: firms are making the skills workers acquire on the job more specific to their own production processes, reducing their value to competing employers. This strategy is especially attractive where new technologies make firm-specific skills cheaper to implement. Because rival firms have less incentive to bid for workers whose skills they cannot use, incumbent firms retain more of the productivity gains generated during the employment relationship, while workers face weaker outside options and move less often. To study this mechanism, I develop an equilibrium search model in which firms choose how strongly their production processes favor firm-specific skills, trading off the cost of implementing them against the wage increases they can avoid. I estimate the model using matched employer–employee data linked to firm balance sheets from France, comparing the 1990s with the 2010s. The model quantifies how much of the decline in mobility is driven by firms' choices of skill specificity and identifies a policy-relevant externality: firms do not internalize the costs their choices impose on other employers by making recruitment more difficult. As a result, skills may be less portable than is socially efficient, reducing both output and worker mobility. I use the model to quantify these losses and assess the potential gains from policies that promote skill portability, such as certification incentives.