Zirui Wang

Retail mortgage rates vary substantially among similar borrowers. Mortgage brokers promise to shop across lenders, yet brokered loans have average rates and rate dispersion similar to retail loans. Using novel mortgage origination data linking borrowers, lenders, and brokers in the U.S., I study why brokers fail to reduce these price differences. I document wide, persistent differences in broker pricing: some consistently deliver rates above retail benchmarks. An IV exploiting relative proximity to broker branches shows that working with a high-cost broker raises mortgage rates by 10–12 basis points. I examine brokers’ lender choices and find that some systematically place borrowers with more expensive lenders, even when cheaper alternatives are available. These choices are associated with higher compensation paid by lenders to brokers, consistent with steering. Steering persists within brokers over time, and borrowers working with brokers with stronger past steering practices receive higher rates. Overall, the findings show how conflicts of interest can undermine the benefits of delegated search and raise household borrowing costs.

We find that crypto AML policies targeting services, such as mixers and exchanges, are more effective than those targeting users: Tornado Cash sanctions resulted in an 80% decline in illicit inflows, while sanctions on individuals often come late and criminals keep using related, non-sanctioned wallets.

Commercial property insurance costs have grown more than 15% annually since 2019; the burden is increasingly shifting from tenants to owners, and high-risk properties are reallocating toward larger owners with pricing advantages.