Publications

Dwyer, Gerald P., Gilevska, B., Nieto, Maria, J. and Samartín, M. 2023. The effects of the ECB’s unconventional monetary policies from 2011 to 2018 on banking assets. Journal of International Financial Markets, Institutions and Money. Vol(87), 101800.

https://doi.org/10.1016/j.intfin.2023.101800.

(https://www.sciencedirect.com/science/article/pii/S1042443123000689).

Working Papers


CMBS and Correlated Investment Exposure: Empirical Evidence 

with Rebel A. Cole


We use the U.S. small commercial banks to analyze the impact of portfolio holdings of commercial mortgage-backed securities (CMBS) on a bank’s correlated investment exposure. Correlated investment exposure measures a bank's systemic risk-taking arising from common asset holdings. We examine the period before the global financial crisis when this market uniquely experienced exceptional growth, and a drop in bank capital requirements from 8% to 1.6% to fund such holdings. We employ quantile regression to show that when analyzed horizontally, banks’ holdings of commercial mortgage-backed securities contributed to a latent buildup of correlated investment exposure across banks. For bank management, our findings indicate that the risk attributes of commercial mortgage-backed securities are different for banks holding these securities than for banks issuing them. For banking regulation, we document how low capital requirements can inadvertently promote homogeneity in asset allocation, thereby heightening systemic vulnerability within the banking sector. 



Asset commonality on bank balance sheets: Implications for bank liquidity 

with Rebel A. Cole and Esteban Hernandez


We examine asset commonality on bank balance sheets and its implications for bank liquidity. We measure asset commonality as asset portfolios overlap across banks. We use COVID-19 shock as an exogenous event to examine whether a higher asset commonality coefficient impacts a bank’s liquidity adversely. We run logic regression models to show that banks with high asset commonality coefficients exhibited a significant drop in bank liquidity following the initial COVID-19 shock. In addition, we examine specifically the role of marketable assets in banks’ asset commonality and its implications for bank liquidity. For this purpose, we measure asset commonality as liquidity-weighted portfolio overlap. We show that banks that have portfolios with larger holdings of marketable assets have a stronger impact on bank liquidity.

Our main findings indicate that marketable assets actually, expose banks to asset commonality and impose uncertainty with respect to banks’ liquidity. Thus, our findings raise the concern of defining and controlling the circumstances when we can categorize certain assets as liquid assets on banks’ balance sheets in an advanced and sophisticated financial system. Particularly, our findings suggest that asset commonality should represent a determinant of a bank's liquidity. 





Green transition, ESG lending, and bank systemic risk 

with Pedro Cuadros Solas


We aim to improve our understanding of whether green and ESG lending represents a relevant factor for banks’ contribution to systemic risk in any significant way.  There is a wide recognition that the inclusion of green and sustainability standards into bank lending, trading, and investment practices is critical for achieving the core mandates of International Financial Organizations and agreements, such as the International Monetary Fund, the World Bank, and Paris agreement. Therefore, as key financial intermediaries, banks have taken the lead in channeling capital toward firms’ green and environmentally sustainable transformation. Taking into account how green and ESG lending have shaped banks’ loan portfolios, the important question that arises is what the implications for individual bank’s contribution to systemic risk are. Given banks’ common exposure to the environmental and sustainability sectors through the syndicated lending channel, we expect that there will be a substantial increase in banks’ overall contribution to systemic risk through the balance sheet propagation channel. Therefore, we focus on providing two important findings in this study. First, banks' large exposures to the environmental sectors through their balance sheets imply that green and ESG lending generate bank interconnectedness. Next, many theoretical works showed that interconnectedness can increase systemic risk through various forms of financial contagion because of the common exposures in times of crises (Allen et al., 2009, Ibragimov et al. 2011, Wagner, 2010, 2011). This implies that green and ESG lending increases banks’ contribution to systemic risk.  Hence, in this research study, we quantify the relationship between banks’ green and ESG lending and bank interconnectedness and systemic risk. 





The role of creditor protection in bank lending: International evidence 

with Rebel A. Cole


In this paper, we focus on how legal origin and creditor protection affect bank lending. Consistent with the “law-and-finance” literature and the “power” theory of credit, we hypothesize that the loan-to-asset ratio of a bank is a function of its country’s legal tradition and how well that country’s legal and judicial systems protect creditors. We expect credit from financial intermediaries as a share of assets to be higher in countries of English common law legal origin and lower in countries of French civil law legal origin. Also, better creditor protection in the form of stronger legal rights or more efficient judicial enforcement has the effect of reducing the expected loss rate on the bad-loan portfolio, which should lead to a higher loan-to-asset ratio.

 

Most studies have analyzed country-level data, usually focusing on how investor protection affects the amount of private-sector credit, which King and Levine (1993) and many other studies have linked to future economic growth. We base our analysis on bank-level financial data from developed countries for 2012-2020. We focus on this period because we want to avoid contamination from the 2008 global financial crisis and the Covid-19 effects. Our primary contribution to the literature is new evidence from bank-level data of a bank-lending channel by which better legal protection, leads to more credit and, consequently, to better financial sector development. With better judicial enforcement, bankers increase the portion of their asset portfolios allocated to loans. In aggregate, this should lead to higher levels of private sector credit, which the “finance and growth” literature has shown to be positively related to economic growth.




Work in progress


The role of AI in screening standards for consumer and business loans 

with Selma Izadi


This paper investigates the role of artificial intelligence (AI) in screening standards in the context of consumer and business loans. It has been widely accepted that AI is rapidly transforming the banking industry by improving efficiency, enhancing customer experience, and reducing risk. Banks especially acknowledge the role of AI in enhancing loans and mortgage processing. It accelerates document review and decision-making, improving turnaround times for such processes. Particularly, banks find AI useful in credit risk assessment across two steps: credit scoring and real-time risk monitoring. AI improves the credit scoring analysis by allowing access to non-traditional data (e.g., mobile usage, social media behavior) to assess creditworthiness. Next, AI allows real–time risk monitoring by adjusting lending strategies dynamically based on customer behavior and market shifts. In this context, we examine how efficient the AI is in screening borrowers for consumer and business loans. We focus our analysis on the U.S. lending market. We exploit a specific rule of thumb in this lending market based on the summary measure of borrower credit quality known as the FICO score. This rule of thumb allows for exogenous variation when comparing loans with similar characteristics. In our empirical setting, we compare loans processed by AI models vs. loans processed by humans. We measure how much the AI improves two issues in the loan screening process. First, we examine whether AI allows for higher loan issuance relative to human processing of loans. Next, we focus on whether AI mitigates some information asymmetry related to the borrowers’ quality. That is, is AI efficient in distinguishing between good and bad borrowers? Finally, the purpose of the study is to provide evidence on whether AI could enhance loan issuance for good borrowers.