作者: DAVID K. HORNE
关键词: Mortgage loan 、 Statistical model 、 Specification 、 Economics 、 Race (biology) 、 Actuarial science 、 Financial services 、 Highly sensitive
摘要: This study examines the role of race in home mortgage lending by investigating sensitivity estimates to variations model specification. I compare parameter based on a statistical utilized Federal Reserve Bank Boston, using subset data that corresponds FDIC-supervised institutions, with obtained from several alternative specified reflect information reviews loan application files. Estimates effect are shown be highly sensitive assumptions underlie model; minor modifications specification sufficient eliminate effect. The empirical results suggest models used evaluate impact may not provide reliable about bias.