The paper develops a residential supply function for approved fixed-rate mo
rtgages in US commercial banks as a first step to explain differences in or
igination patterns among groups of borrowers. It models the lender's decisi
on to offer the borrower a risk-adjusted loan bundle relative to the terms
on the credit application, The model includes multiple dependent variables
that are risk-adjusted simultaneously to reflect accurately the loan office
r's decision. Canonical correlation factor analysis is used to capture the
lender's simultaneous decision, The loan-price ratio and contract interest
rate are the most important variables in the lender supply function.