Theses and Dissertations

Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Business Administration

First Advisor

Dr. Diego Escobari

Second Advisor

Dr. Andre Mollick

Third Advisor

Dr. Elena Quercioli


An influential literature has applied regression-discontinuity (RD) analysis to investigate the effects of government housing policies. This literature has failed to identify substantial effects of these policies, but may also suffer from various biases and limitations. This dissertation carefully examines what, if any, issues face RD analysis in this context. Chapter II considers the Affordable Housing Goals (AHGs) and Community Reinvestment Act (CRA), affordable housing policies that compel Fannie Mae and Freddie Mac (F&F) and depository institutions, respectively, to increase access to mortgage credit for borrowers perceived by some as “underserved” by the mortgage market. Chapter III considers the overall effects of F&F. For Chapter II, the data covers over 108 million mortgage applications submitted across the U.S. over two decades while Chapter III uses a smaller but more detailed sample of mortgages. Both chapters rely on methods which are the state-of-the-art in RD methodology. The finding of Chapter II is that RD analysis is technically unidentified in this context. This likely explains why substantial effects of the AHGs and CRA do not appear and it is also consistent with the view that the policies contributed to the subprime crisis by compelling F&F and depository institutions to enter or expand in the subprime mortgage market. In a similar vein, Chapter III provides striking evidence that borrowers’ ability to manipulate their loan amounts threatens RD analysis in this context. It is shown that borrowers manipulate aggressively when there is a downturn in the housing market or national economy while many borrowers do not bother with manipulation when the market and economy are doing well. This is consistent with the idea that most borrowers are highly dependent on F&F and this dependency becomes stronger during downturns. However, some borrowers are able to obtain reasonably price mortgage credit from the nonconforming mortgage market during times of growth. These borrowers are shown to be the very borrowers for whom RD analysis estimates the effect of F&F. It is thus no wonder that these estimates are generally zero. The conclusion is that RD analysis cannot reliably estimate the effects of government housing policies.


Copyright 2017 Alejandro Rojas. All Rights Reserved.