We use unique data from BHC-affiliated securities dealers in the United States to study the use of collateral in bilateral retirement and securities lending transactions. This is the first document that provides stylized facts about this market and shows, among other things, that most trades are backed by U.S. Treasury bonds and stocks and have either an overnight lifespan or an open duration. We also focus on how haircuts are determined in this market and highlight the differences between asset classes. Finally, we document a negative correlation between haircut and interest rates, a forecast that is consistent with a multitude of theories; However, this correlation is quite weak. We use unique data from securities dealers affiliated with U.S. banks to study the use of collateral in bilateral retirement and securities lending transactions. The use of collateral by operators varies considerably by asset class: for the United States. When it comes to transactions in treasury securities, we find that haircuts are large enough to offer comprehensive protection against defaults, while this is generally not the case for stock market transactions. While most stocks in our sample are associated with a single haircut, most U.S. Treasury bonds are associated with more than one haircut. We refer to these results to the implications of the zero-value-at-risk characteristic found in guarantee theories as an enforcement mechanism, and we show that the data do not support these implications.

We then turn to negative selection models, which predict a negative relationship between haircuts and interest rates, based on the use of collateral as a screening mechanism. We only find this relationship negative for trades where securities dealers receive U.S. Treasury bonds and provide cash. Federal Reserve Bank of New York Research Paper Series. Access to this document is limited, so you can search for another version below or another version of it. . Subscribe to this paid journal for further articles on the topic The authors thank Matteo Aquilina, Jun Kyung Auh, Ana Fostel and Gregory Phelan, as well as seminar and conference participants for their comments and suggestions. In addition, we thank Jacob Adenbaum, Ocean Dalton, Aaron Plesset and Noah Zinsmeister for their excellent support of the research…