The Community Reinvestment Act (CRA) encourages lenders to make mortgage loans to certain classes of borrowers. However, the law does not apply to all lenders, and lenders do not necessarily receive credit for all loans made to borrowers of a particular class. We use this variation to test whether or not CRA-affected lenders cut interest rates to CRA-eligible borrowers; in other words, we test for the presence of a regulation-driven subsidy. Our theory suggests that loans made by commercial banks and savings associations (``relationship lenders'') and mortgage companies (``transaction lenders'') will differ from one another depending on borrower risk and homeownership benefits. Empirically, we find that CRA-eligible loans at CRA-affected institutions do carry lower mortgage spreads compared with other loans at the same institution. However, once we control for risk and benefit effects suggested by our theory, these differences in mortgage spreads become economically and statistically insignificant.