We study how the 1990s adoption of now widely-used automated mortgage underwriting systems affected credit, house prices and their comovement across locations. The effects go well beyond processing improvements. By implementing more complex, statistically-informed lending rules, the systems allowed households to borrow more, pushing up house prices. Furthermore, by transmitting a common set of credit standards across lenders, the new technology increased house price synchronization. Together, our results illustrate how new lending technology can generate systematic credit supply shocks, influencing house prices and increasing market interconnectedness.

U.S.  homeowners' insurance premiums have increased drastically, widely attributed to climate risk. Using household-level microdata and two novel instruments, we document three findings. First, households facing larger premium increases are more likely to relocate, moving to areas with lower climate risk and insurance costs. Homes with larger increases are more often acquired by investors, who obtain lower premiums. Second, higher premiums increase mortgage delinquencies. Third, the relocation effect is stronger among less financially constrained households, while the delinquency effect concentrates among more constrained ones. Rising insurance costs can threaten household financial resilience, while relocation offers a costly mechanism of adaptation.

I show that policies restricting the ratio of mortgage payments to income are particularly costly for self-employed workers. I identify these effects using local variation in exposure to the Ability-to-Repay and Qualified Mortgage rule. By limiting mortgage payment size as a share of stable, documented income, the rule disproportionately reduced credit to self-employed workers and led to a decline in self-employment and small business employment. The employment response is concentrated in new businesses and in industries where start-up costs are low enough to be financed primarily with mortgage credit.

I measure the effect of mortgage leverage restrictions on house prices using a change in eligibility requirements imposed by Fannie Mae and Freddie Mac. I use a machine-learning approach to document an unannounced divergence in Fannie and Freddie’s automated underwriting algorithms. This led to tighter lending standards and lower house prices in places where local lenders had pre-existing relationships with Freddie. The empirical response is close to a simple theoretical benchmark with no substitution, suggesting the effect of a large market participant’s lending standards on house prices is close to the theoretical upper bound.

Conducting a new survey on the Nielsen Consumer panel we examine the impact of working from home (WFH) on shopping, finding three key results. First, WFH changed shopping modes, increasing online shopping and the fraction of trips to stores on weekdays. Second, WFH increased total spending through increased product quantity and range, tilting expenditure towards food and general merchandise and away from health and beauty products. Finally, WFH increased prices paid by shoppers and lowered price elasticities due to a move towards higher-cost products and lower deal usage. The change in prices paid is concentrated among married households, driven by a redistribution of shopping responsibility within the household, especially when the new remote worker is male. We find that remote workers engage in more shopping trips but fewer minutes of shopping. Our results suggest that shifting to remote work increases prices paid for groceries by about 1%, highlighting how WFH is impacting households, retail and inflation.

Households tend to hold substantial amounts of non-financial assets in the form of consumer goods inventories that are unobserved by traditional measures of wealth, about $725 on average for products covered by our sample. Such holdings can eclipse total financial assets among households in the lowest income quintile. Households can obtain significant financial returns from strategically shopping and managing these inventories. In addition, they choose to maintain liquid savings—household working capital—not just for precautionary motives but also to support this inventory management. We demonstrate that households earn high marginal returns from investing in household working capital, well above 20% at low levels of inventory, though these marginal returns decline rapidly as inventory increases. Nevertheless, average returns from inventory management are high—about 50% for the typical household—and affect household portfolio returns substantially for all but the top income and asset quintiles. We provide evidence from scanner and survey data that supports this conclusion. For many households, working capital is therefore an important asset class that has been largely ignored by the household finance literature, and inventory management provides them with an alternative to investing in risky financial markets at low levels of liquid wealth.

Using comprehensive high-frequency state and local sales tax data, we show that shopping behavior responds strongly to changes in sales tax rates. Even though sales taxes are not observed in posted prices and have a wide range of rates and exemptions, consumers adjust in many dimensions. They stock up on storable goods before taxes rise and increase online and cross-border shopping in both the short and long run. The difference between short- and long-run spending responses has important implications for the efficacy of using sales taxes for counter-cyclical policy and for the design of an optimal tax framework. Interestingly, households adjust spending similarly for both taxable and tax-exempt goods. We embed an inventory problem into a continuous-time consumption-savings model and demonstrate that this behavior is optimal in the presence of shopping trip fixed costs. The model successfully matches estimated short-run and long-run tax elasticities. We provide additional evidence in favor of this new shopping-complementarity mechanism.

This paper studies how credit markets respond to policy constraints on household leverage. Exploiting a sharp policy-induced discontinuity in the cost of originating certain high-leverage mortgages, we study how the Dodd-Frank “Ability-to-Repay” rule affected the price and availability of credit in the U.S. mortgage market. Our estimates show that the policy had only moderate effects on prices, increasing interest rates on affected loans by 10-15 basis points. The effect on quantities, however, was significantly larger; we estimate that the policy eliminated 15 percent of the affected market completely and reduced leverage for another 20 percent of remaining borrowers. This reduction in quantities is much greater than would be implied by plausible demand elasticities and indicates that lenders responded to the policy not only by raising prices but also by exiting the regulated portion of the market. Heterogeneity in the quantity response across lenders suggests that agency costs may have been one particularly important market friction contributing to the large overall effect as the fall in lending was substantially larger among lenders relying on third parties to originate loans. Finally, while the policy succeeded in reducing leverage, our estimates suggest this effect would have only slightly reduced aggregate default rates during the housing crisis.