We study the impact of hedge funds' bargaining power on banks' haircut policies in secured lending transactions. We observe that on the same day, for identical collateral, and under identical repo contracts, banks require significantly lower haircuts from hedge funds with greater bargaining power, even when controlling for their probability of default. This effect is further confirmed by plausible exogenous variation in hedge funds' bargaining power, stemming from Credit Suisse's withdrawal from the prime brokerage business. Furthermore, our findings reveal that higher bargaining power of hedge funds substantially elevates the risk of insufficient haircuts according to standard value-at-risk models, in particular for collateral eligible for monetary policy operations.

This paper investigates the impact of ESG short-selling activity in light of the recent ESG backlash and the rise of passive investment. We postulate that ESG short-selling activity varies depending on whether stocks are included in a blue-chip equity index, influenced by the growing prominence of passive investment strategies. Using short-selling data from IHS Markit and ESG ratings from Refinitiv for U.S. equities from January 2016 to December 2022, we find that higher-overpriced ESG stocks excluded from the MSCI USA Index are no longer immune to short sellers. Our results also reveal that the securities lending market

supply is not a constraint for short sellers when stocks are excluded from the index,

particularly in the context of ESG considerations. Finally, we find that borrowing costs for higher-ESG stocks are higher for those outside the index, while the opposite is true for stocks included in the index.

This research presents a comprehensive framework for optimizing Volume Weighted Average Price (VWAP) execution in cryptocurrency markets using deep learning approaches. Through three interconnected studies, we demonstrate how moving beyond traditional volume curve prediction can enhance VWAP execution performance. First, we show how deep learning's automatic differentiation capabilities can directly optimize VWAP execution by minimizing either absolute or quadratic deviations from market VWAP. Our initial static model, implemented using a Temporal Linear Network architecture, consistently outperforms traditional volume-prediction approaches across multiple cryptocurrencies. Building on these results, we develop a dynamic execution framework utilizing recurrent neural networks that can adapt to changing market conditions during order execution. This dynamic approach further improves performance by incorporating real-time market feedback into the execution strategy. Finally, we introduce a novel multi-asset learning approach using signature-enhanced transformers, which enables efficient model deployment across multiple assets while maintaining superior performance. This architecture, which combines attention mechanisms with path signatures, demonstrates consistent improvement over both asset-specific and globally-fitted models for both previously seen and unseen assets. Our research not only advances the theoretical understanding of VWAP execution but also provides practical implementations that significantly improve trading efficiency in cryptocurrency markets.