I provide new evidence that large and small banks have different external financing costs, which generates cross sectional variation in a deposits market pricing power channel of monetary policy transmission. I do so by exploiting a natural experiment using anti-trust related bank branch divestitures. In these divestitures branches are transferred from large to small banks such that the branches are largely preserved and market structure remains unchanged. Consistent with the existence of capital market imperfections at small banks, I find that – holding market concentration constant – lending declines in areas local to branches newly owned by small banks, and deposit rates increase by more, when interest rates increase. My results are confirmed with geographically granular tests utilizing the entire sample of data, and indicate that financing frictions still matter for the transmission of monetary policy through banks.

In this paper I document two new facts. First, bank net-interest margins (NIM) are insensitive to the short rate on average but this masks substantial heterogeneity in the cross section. I find cross sectional variation ranging from a -30bp to +40bp change in one quarter NIM after a 100bp change in rates, which compounds over the following four quarters. I find that this cross sectional heterogeneity is driven by differences in bank business models: banks with large negative NIM sensitivity to rate changes hold more residential mortgages, and rely more on interest bearing deposits whereas banks with large positive NIM sensitivity hold more typically shorter term commercial and industrial loans and rely more on non-interest bearing deposits. Second, bank NIM sensitivity to rate changes is muted by offsets in non-interest expenses, reducing ROA sensitivity by approximately 30% on average. The positive relationship between net-interest income and non-interest expenses is not consistent with investment sensitivity to cash-flows as a result of costly external funds, but is somewhat consistent with investment sensitivity to profitability. However I also find that banks pass through NIM declines on the upside but not the downside, and the intensity of the relationship between net-interest income and non-interest expenses substantially increases closer to fiscal year end. These adjustments are consistent with the management of expenses in order to prop up earnings to meet executive performance targets, and have the side effect of reducing earnings sensitivity to rate changes.

We examine the effects of the Mortgage Electronic Registration System, or MERS, on mortgage origination volumes and foreclosure rates prior to the Great Recession. MERS was introduced in the late 1990s and significantly reduced the cost and time associated with secondary loan sales. Using novel data from the Massachusetts Registry of Deeds, we show that the introduction of MERS led to an expansion in credit supply that was primarily fueled by non-bank lenders originating loans to low-income borrowers. We also find that foreclosure rates were higher on these loans. Our paper provides a new explanation for why credit supply increased prior to the 2008 financial crisis and why supply increases were larger in low-income areas.