Multiplex network analysis of the UK OTC derivatives market with Marco Bardoscia and Ginestra Bianconi (2019) - International Journal of Finance & Economics (forthcoming)
Abstract: In this paper, we analyse the network of exposures constructed by using the UK trade repository data for three different categories of contracts: interest rate, credit, and foreign exchange derivatives. We study how liquidity shocks related to variation margins propagate across the network and translate into payment deficiencies. A key finding of the paper is that, in extreme theoretical scenarios where liquidity buffers are small, a handful of institutions may experience significant spillover effects due to the directionality of their portfolios. Additionally, we show that a variant of a recently introduced centrality measure - Functional Multiplex PageRank - can be used as a proxy of the vulnerability of financial institutions, outperforming in this respect the commonly used eigenvector centrality.
Systemic Illiquidity in the Interbank Network with Sam Langfield, Zijun Liu and Tomohiro Ota (2019) - Quantitative Finance
Abstract: We study systemic illiquidity using a unique dataset on banks' daily cash flows, short-term interbank funding and liquid asset buffers. Failure to roll-over short-term funding or repay obligations when they fall due generates an externality in the form of systemic illiquidity. We simulate a model in which systemic illiquidity propagates in the interbank funding network over multiple days. In this setting, systemic illiquidity is minimised by a macroprudential policy that skews the distribution of liquid assets towards banks that are important in the network.
The impact of de-tiering in the UK's large value payment system with Evangelos Benos and Pedro Gurrola-Perez (2017) - Journal of Financial Market Infrastructures
Abstract: Large-value payment systems (LVPS) often have a tiered structure where only a limited number of banks have direct access to these systems and every other institution accesses the system through agency arrangements with the direct participants. As such, a high degree of tiering is often perceived as being associated with credit and operational risks. In this paper we use data around five recent de-tiering events in the United Kingdom's LVPS (CHAPS), to assess the impact of de-tiering on these risks as well as on liquidity usage. We find that the impact of de-tiering is largest on credit risk where average intraday exposures, between first and second-tier banks, drop by anywhere between 0.3 billion and 1.5 billion pounds per bank, while the cost of insuring against losses arising from these exposures, drops by about 4 million to 19 million pounds per bank, per year. On the other hand, the impact of these de-tiering events on operational risk and liquidity usage appears to be economically small.
The small bank failures of the early 1990s: another story of boom and bust with Kushal Balluck, Artus Galiay, and Glenn Hoggarth - Bank of England Quarterly Bulletin
Abstract: The article describes in non-technical language how the deregulation of the UK retail banking system heavily affected the banking system's stability. As a matter of interest, it should be noted that banking system credit losses in the United Kingdom in the early 1990s were over three times higher than they were in the recent financial crisis. The Bank of England is now in a better position to guard against many of the vulnerabilities that led to the small banks crisis. However, history suggests that regulators should continually look for early warning signs of heightened risk in the financial system, such as rapid credit growth, a decline in underwriting standards and large shifts in business models.
Full Payment Algorithm with Marco Bardoscia, Nicholas Vause and Michael Yoganayagam (2019)
Abstract: Clearing payments between firms are usually computed under the assumption that firms that cannot fully meet their obligations exhaust their cash resources to make partial payments. In practice, however, firms might wait and see whether they receive payments that would help them to meet their obligations in full before making any payments themselves, thereby avoiding partial payments. In this paper, we model this situation by introducing the Full Payment Algorithm (FPA). We fully characterize the steady state of the FPA, and we prove necessary and sufficient conditions under which it is equivalent to the widely used Eisenberg and Noe model.
The impact of the leverage ratio on client clearing with Jonathan Acosta-Smith and Francesc Rodriguez-Tous (2018)
Price awarded at the 27th finance forum: Best paper on Regulation
Abstract: As part of the post-crisis regulatory reform, many interest-rate derivative transactions are required to be centrally cleared. Nevertheless, the treatment of this type of transaction under the leverage ratio (LR) requirement does not allow for the use of initial margin to reduce the exposure, thereby increasing capital costs. As a result, LR affected clearing member banks may be more reluctant to provide central clearing services to clients given this additional cost. This in turn can prevent some real economy firms from hedging their risks. We analyse whether this is the case by exploiting detailed confidential transaction and portfolio level data as well as the introduction and posterior tightening of the LR in the UK in a diff-in-diff framework. Our results suggest that the LR had a disincentivising effect on client clearing, both in terms of daily transactions as well as the number of clients, but this impact seems to be driven by a reduced willingness to take on new clients.
Central counterparty auction design with Xin Li and Daniel Marszalec (2017)
Abstract: We analyze the role of auctions in managing the default of a clearing member in a generic central counterparty (CCP). We first consider three established alternative sealed bid auction formats in which clearing members simultaneously submit bids for a defaulting clearing member's portfolio: first price without penalty, first price with penalty, and first price with budget constraints. Under our assumptions regarding bidders' behavior, although the revenue of the portfolio by the CCP might be the same for these auction formats mentioned above, there could be significant differences in the externalities arising from each of them. Additionally, this paper considers how mechanisms to incentivize competitive bidding could, in some circumstances, have adverse implications for financial stability by inefficiently distributing losses to surviving clearing members. In response to these potential adverse implications, we propose a fourth auction type - second price with loss sharing - which takes into account a bidder's consideration that may bear part of the CCP's losses.
How Trade Matching Forms in the Credit Default Swap Market with Jun Sung Kim, Bonsoo Koo and Zijun Liu (2015)
Abstract: We investigate how the pairing of dealers and customers in credit default swap (CDS) transactions is influenced by the participants characteristics. Using data from the Depository Trust & Clearing Corporation on trade matching decisions in the UK CDS market, we employ a matching/network formation framework to uncover participant preferences. We also extend the reference-entity-specific approach by applying a panel fixed effects model to attain, in combination, a comprehensive understanding of the formation of the UK CDS network. The matching/network formation approach supports the theoretical prediction that the UK CDS market is congregating on a small number of market participants that are able to offer lower transaction costs due to economies of scale. Our reference-entity-specific analysis finds that network formation is an outcome of the interplay between hedging and speculative purposes in the UK CDS market.
Optimization under Model Uncertainty Job Market Paper (2014)
Abstract: This study proposes a novel methodology to deal with model uncertainty in forecasting stock returns. My main interest here is to overcome the tendency of Bayesian Model Averaging to give all of the weight to a single model. A potential solution of this problem is to capture the nature of the underlying data generating process by sampling from the space of possible combinations. After presenting the methodology in detail, I show that my approach increases the accuracy of out-of-sample forecasting. Moreover, I investigate the impact of improved forecasting on portfolio performances.
Abstract: Over the years researchers have attempted to develop asset allocation models to aid decision making in portfolio selection, based upon a statistical or a heuristic approach. Optimization of asset allocation is often complex and nonlinear with many local optima. For this reason, searching the global solution by analytical methods is computationally expensive and ineffective. In comparison with other local search algorithms, Genetic Algorithms (GA) assures a higher chance of reaching a global optimum by starting with multiple random search points and considering several candidate solutions simultaneously.
Derivatives stress simulation using trade repository data with Marco Bardoscia, Nicholas Vause, and Michael Yoganayagam (2017)
CCPs' waterfall implications for moral hazard with Cristina Picillo (2017)