How Quantitative Easing Works: Evidence on the Refinancing Channel
76 Pages Posted: 9 Dec 2015 Last revised: 6 Dec 2016
Date Written: September 26, 2016
Despite massive large-scale asset purchases (LSAPs) by central banks around the world since the global financial crisis, there is a lack of empirical evidence on whether and how these programs affect the real economy. Using rich borrower-linked mortgage-market data, we document that there is a “flypaper effect” of LSAPs, where the transmission of unconventional monetary policy to interest rates and (more importantly) origination volumes depends crucially on the assets purchased and degree of segmentation in the market. For example, QE1, which involved significant purchases of GSE-guaranteed mortgages, increased GSE-eligible mortgage originations significantly more than the origination of GSE-ineligible mortgages. In contrast, QE2's focus on purchasing Treasuries did not have such differential effects. We find that the Fed's purchase of MBS (rather than exclusively Treasuries) during QE1 resulted in an additional $600 billion of refinancing, substantially reducing interest payments for refinancing households, leading to a boom in equity extraction, and increasing consumption by an additional $76 billion. This de facto allocation of credit across mortgage market segments, combined with sharp bunching around GSE eligibility cutoffs, establishes an important complementarity between monetary policy and macroprudential housing policy. Our counterfactual simulations estimate that relaxing GSE eligibility requirements would have significantly increased refinancing activity in response to QE1, including a 20% increase in equity extraction by households. Overall, our results imply that central banks could most effectively provide unconventional monetary stimulus by supporting the origination of debt that would not be originated otherwise.
Keywords: Monetary policy, MBS, quantitative easing, LSAP, Refinancing, deleveraging, HARP, GSE
JEL Classification: E20, E30, E51, G28
Suggested Citation: Suggested Citation