We study systemic risk in a network model of the interbank market where the asset returns of the banks in the network are correlated. In this way we can study the interaction of two important channels for systemic risk (correlation of asset returns and contagion due direct financial linkages). We carry out a simulation study that determins the probability of a systemic crisis in the banking network as a function of both the asset correlation, and the connectivity and structure of the financial network. An important observation is the fact that the relation between asset correlation and the probalility of a systemic crisis is hump-sharped; in particular, lowering the correlation between the assets returns of different banks does not always imply a lower probability of a systemic crisis.

Management of systemic risk in financial markets is traditionally associated with setting (higher) capital requirements for market participants. There are indications that while equity ratios have been increased massively since the financial crisis, systemic risk levels might not have lowered, but even increased (see ECB data 1 ; SRISK time series 2 ). It has been shown that systemic risk is to a large extent related to the underlying network topology of financial exposures. A natural question arising is how much systemic risk can be eliminated by optimally rearranging these networks and without increasing capital requirements. Overlapping portfolios with minimized systemic risk which provide the same market functionality as empir- ical ones have been studied by Pichler et al. (2018). Here we propose a similar method for direct exposure networks, and apply it to cross-sectional interbank loan networks, consisting of 10 quarterly observations of the Austrian interbank market. We show that the suggested framework rearranges the network topol- ogy, such that systemic risk is reduced by a factor of approximately 3.5, and leaves the relevant economic features of the optimized network and its agents unchanged. The presented optimization procedure is not intended to actually re-configure interbank markets, but to demonstrate the huge potential for systemic risk management through rearranging exposure networks, in contrast to increasing capital requirements that were shown to have only marginal effects on systemic risk (Poledna et al., 2017). Ways to actually incentivize a self-organized formation toward optimal network configurations were introduced in Thurner and Poledna (2013) and Poledna and Thurner (2016). For regulatory policies concerning financial market stability the knowledge of minimal systemic risk for a given economic environment can serve as a benchmark for monitoring actual systemic risk in markets.

Thesis advisor: Matthew O. Jackson / In this dissertation, I focus on networks in macroeconomics and finance. In Chapter 1, I develop a theoretical model of rescue of distressed financial institutions. I study rescues in a coalition formation framework, which provides new insights into the financial contagion and stability and rescue of systemically important financial institutions. The findings show that the levels of negative shock, bankruptcy costs, interbank obligations of each financial firm and the topology of the interbank network all together determine financial firms’ contributions in rescues, where government assistance in rescues is not required in certain types of network structures. In Chapter 2, which is a joint work with Matthew O. Jackson, we study the impacts of sector level technological changes on wage inequality and GDP growth in production networks. Our results show that the macroeconomic implications of sector level technological changes depend on additional factors than the input-output structure such as type of the intermediate good (e.g., substitutes for labor vs complements to labor), task weights in production processes and labor supply. Chapter 1. I model bank rescues in a setting where banks hold each other’s financial instruments creating a network of financial linkages. Costly bankruptcies reduce interbank payments, which creates incentives for rescues by other banks. Accordingly, I analyze the sources of inefficiencies in bank rescues and show that the social welfare is maximized if regulators promote financial networks that are evenly connected (without disconnectedness/clustering) and have intermediate levels of interbank liabilities at bank level. Such networks maximize banks’ total contributions to the rescue of a distressed bank hit by a relatively small negative shock, but also ensure that banks do not fail sequentially like dominos when a bank hit by a large shock does actually fail. The results also provide a rationale for why some systemically important banks were not rescued in 2007-2008. In the model, a social welfare maximizing government assists the rescues designed to prevent the potential contagious failures and maintain financial stability instead of assisting the rescue of a bank that is hit by a large shock. Chapter 2. We study the impact of technological change on wage inequality and GDP growth in production networks. We do this in a simple model that contrasts the effects of changes in intermediate goods that substitute for labor with those that complement labor. Technological changes in intermediate goods that complement labor result in increased GDP and do not change relative wages. Technological changes in intermediate goods that substitute for (low-skilled) labor involve three phases: pre-automation, transition to automation, and post-automation. During the transition phase, technological changes in such intermediate good lead to increased wage inequality and relatively smaller increases in GDP than comparable changes in complementary goods. In addition, our results show that firm-level weights of tasks performed by different types of labor play key roles in macroeconomic network consequences of interconnectedness. / Thesis (PhD) — Boston College, 2018. / Submitted to: Boston College. Graduate School of Arts and Sciences. / Discipline: Economics.

Cette thèse est composée de quatre chapitres.Le premier chapitre est une description préliminaire de la base de données Factset Ownership. Nous en donnons une description statistique et exposons quelques faits stylisés caractérisant notamment la structure du portefeuille des institutions financières et fonds d'investissement, ainsi que la capitalisation boursière des entreprises y étant recensées.Le second chapitre propose une méthode d'évaluation statistique de la similarité entre des paires de portefeuilles d'institutions financières. Une paire statistiquement significative donnant lieu à la création d'un lien de similarité entre ces deux entités, nous sommes en mesure de projeter un réseau à l'origine bi-partite (entre institutions financières et entreprises) en un réseau mono-partite (entre institutions uniquement) afin d'en étudier l'évolution de sa structure au cours du temps. En effet, d'un point de vue économique, il est suspecté que les motifs d'investissements similaires constituent un facteur de risque important de contagion financière pouvant être à l'origine de banqueroutes aux conséquences systémiques significatives.Le troisième chapitre s'intéresse aux comportements collectifs des gestionnaires de fonds d'investissement et, en particulier, à la manière dont la structure du portefeuille de ces fonds prend en compte, en moyenne, de façon optimale les frais de transaction en présence de faibles contraintes d'investissements. Ce phénomène où, dans de nombreuses situations, la médiane ou la moyenne des estimations d'un groupe de personnes est étonnamment proche de la valeur réelle, est connu sous le nom de sagesse de la foule.Le quatrième chapitre est consacré à l'étude simultanée de données de marché. Nous utilisons plus de 6.7 milliards de trades de la base de données Thomson-Reuters Tick History, et de données de portefeuille de la base FactSet Ownership. Nous étudions la dynamique tick-à-tick du carnet d'ordres ainsi que l'action aggrégée, c'est-à-dire sur une échelle de temps bien plus grande, des fonds d'investissement. Nous montrons notamment que la mémoire longue du signe des ordres au marché est bien plus courte en présence de l'action, absolue ou directionnelle, des fonds d'investissement. Réciproquement nous expliquons dans quelle mesure une action caractérisée par une mémoire faible est sujette à du trading directionnel provenant de l'action des fonds d'investissement. / The thesis is divided into four parts.Part I introduces and provides a technical description of the FactSet Ownership dataset together with some preliminary statistics and a set of stylized facts emerging from the portfolio structure of large financial institutions, and from the capitalization of recorded securities.Part II proposes a method to assess the statistical significance of the overlap between pairs of heterogeneously diversified portfolios. This method is then applied to public assets ownership data reported by financial institutions in order to infer statistically robust links between the portfolios of financial institutions based on similar patterns of investment. From an economic point of view, it is suspected that the overlapping holding of financial institution is an important channel for financial contagion with the potential to trigger fire sales and thus severe losses at a systemic level.Part III investigates the collective behaviour of fund manager and, in particular, how the average portfolio structure of institutional investors optimally accounts for transactions costs when investment constraints are weak. The collective ability of a crowd to accurately estimate an unknown quantity is known as the Wisdom of the Crowd. In many situation, the median or average estimate of a group of unrelated individuals is surprisingly close to the true value.In Part IV, we use more than 6.7 billions of trades from the Thomson-Reuters Tick History database and the ownership data from FactSet. We show how the tick-by-tick dynamics of limit order book data depends on the aggregate actions of large funds acting on much larger time scale. In particular, we find that the well-established long memory of marker order signs is markedly weaker when large investment funds trade in a markedly directional way or when their aggregate participation ratio is large. Conversely, we investigate to what respect an asset with a weak memory experiences direction trading from large funds.

We present the rich research potential at the union of multi agent reinforcement learning (MARL), game theory, and financial graphs. We demonstrate how multiple game theoretic scenarios arise in three node financial graphs with minor modifications. We highlight six scenarios used in this study. We discuss how to setup an environment for MARL training and evaluation. We first investigate individual games and demonstrate that MARL agents consistently learn Nash Equilibrium strategies. We next investigate mixed games and find again that MARL agents learn Nash Equilibrium strategies given sufficient information and incentive (e.g. prosociality). We find introducing a embedding layer in agents deep network improves learned representations and as such, learned strategies, (2) MARL agents can learn a variety of complex strategies, and (3) selfishness improves strategies’ fairness and efficiency. Next we introduce populations and find that (1) pro social members in a population influences the action profile and that (2) complex strategies present in individual scenarios no longer emerge as populations’ portfolio of strategies converge to a main diagonal. We identify two challenges that arises in populations; namely (1) identifying partner’s prosociality and (2) identifying partner’s identity. We study three information settings which supplement agents observation set and find having knowledge of partners prosociality or identity to have negligible impact on how portfolio of strategies converges. / Vi presenterar den rika forskningspotentialen vid unionen av multi-agent förstärkningslärning (MARL), spelteori och finansiella grafer. Vi demonstrerar hur flera spelteoretiska scenarier uppstår i tre nodgrafikgrafer med mindre ändringar. Vi belyser sex scenarier som används i denna studie. Vi diskuterar hur man skapar en miljö för MARL -utbildning och utvärdering. Vi undersöker först enskilda spel och visar att MARL -agenter konsekvent lär sig Nash Equilibrium -strategier. Vi undersöker sedan blandade spel och finner igen att MARL -agenter lär sig Nash Equilibrium -strategier med tillräcklig information och incitament (t.ex. prosocialitet). Vi finner att införandet av ett inbäddande lager i agenternas djupa nätverk förbättrar inlärda representationer och som sådan inlärda strategier, (2) MARL-agenter kan lära sig en mängd komplexa strategier och (3) själviskhet förbättrar strategiernas rättvisa och effektivitet. Därefter introducerar vi populationer och upptäcker att (1) pro sociala medlemmar i en befolkning påverkar åtgärdsprofilen och att (2) komplexa strategier som finns i enskilda scenarier inte längre framkommer när befolkningens portfölj av strategier konvergerar till en huvuddiagonal. Vi identifierar två utmaningar som uppstår i befolkningen; nämligen (1) identifiera partnerns prosocialitet och (2) identifiera partnerns identitet. Vi studerar tre informationsinställningar som kompletterar agents observationsuppsättning och finner att kunskap om partners prosocialitet eller identitet har en försumbar inverkan på hur portföljen av strategier konvergerar.

Passive network reconstruction is the process of learning a structured (networked) representation of a dynamic system through the use of known information about the structure of the system as well as data collected by observing the inputs into a system along with the resultant outputs. This work demonstrates an improvement on an existing network reconstruction algorithm so that the algorithm is capable of consistently and perfectly reconstructing a network when system inputs and outputs are measured without error. This work then extends the improved network reconstruction algorithm so that it functions even in the presence of noise as well as the situation where inputs into the system are unknown. Furthermore, this work demonstrates the capability of the new extended algorithms by reconstructing financial networks from stock market data, and then performing an analysis to understand the vulnerabilities of the reconstructed network to destabilization through localized attacks. The creation of these improved and extended algorithms has opened many theoretical questions, paving the way for future research into network reconstruction.

L'objectif de cette thèse est d'étudier la fragilité financière, c.à.d. la sensibilité du système financier par rapport aux perturbations. La difficulté principale concernant la fragilité financière dans le contexte actuel est la complexité croissante du système financier. piur remédier à ce problème, cette thèse s'inspire des deux courants relativement récents de la recherche économique : l'analyse des réseaux et l'économie comportementale. Les principaux concepts mobilisés sont les mécanismes de diffusion, de cascade et la rationalité limitée. Chapitre 1 étudie les effets des structures locales ds liens, spécifiquement la longueur des cycles transitifs sur la magnitude de la contagion financière. Chapitre 2 propose un modèle dynamique des paniques bancaires, dans lequel les paniques émergent par un mécanisme de cascade des retraits. Le but est de mieux comprendre comment les paniques se forment. Chapitre 3 étudie les paniques bancaires dans un contexte à la fois dynamique et comportemental, avec la présence du mimétisme et l'hétérogénéité des déposants. / This thesis studies financial fragility, i.e. the sensitivity of the financial system with respect to shocks. the main issue of financial fragility in the current context is the increased financial complexity. To address this problem, this study draws inspiration from two relatively recent streams of literature : econopmics of networks and behavioral economics. The main concepts in use are diffusion, cascade and bounded rationality. Chapter 1 studies how petterns of links, specifically, the length of transitive cycles affect the extent of financial contagion. Chapter 2 proposes a dynamic model in which bank runs arise as cascades of withdrawals. The aim is to better understand how bank runs occur. Chapter 3 studies bank runs in a dynamic and behavioral setting, with herding and heterogeneity of depositors.

In the growing multiculturalism of Swedish society, we see a significant amount of citizens of various heritages experience difficulty when venturing into the Swedish labor market. These difficulties have been explained earlier by either cultural or structural factors. The cultural factors state that individuals raised outside of the traditional ―Swedish‖ culture have greater difficulty in understanding the subtle idiomatic properties of the Swedish labor market. On the other hand, the structural factors state that it is the deficiency in Swedish language and education that stand as the main obstacles in finding lucrative professions on the labor market. To determine the structural and/or cultural factors, I have performed field research in restaurants of the Norrköping district. The restaurant is a setting in which several aspects are shown to have influence over the informants‘ decision to become a restaurant owner. These aspects include the labor market, food, family relations, and networks between friends and professional associates. Why have these individuals chosen restaurants? Is it out of economic necessity, or is it a family profession they feel compelled to safeguard? In this thesis, several restaurant owners will answer questions regarding their choice of profession, their prospects on the labor market, the influence of their families, the importance of cuisine as a cultural foundation, and the discrepancies behind social and financial networks among ethnic groups in Sweden.

L'ultima crisi nanziaria ha evidenziato il ruolo decisivo delle connessioni nel mercato interban- cario come canale e strumento ampli catore dei shock nanziari, e di conseguenza del rischio sistemico. In questa tesi presentiamo delle metodologie teoriche ed empiriche per analizzare il potenziale rischio sistemico in una rete bancaria interconnessa. La tesi comprende due saggi sulle reti nanziarie e il rischio sistemico ed e organizzata in due capitoli. Nel capitolo I analizziamo e modelliamo alcune delle complesse interazioni all'interno di una rete nanziaria, con l'obiettivo di approfondire nella interrelazione fra la fragilit a dell'eco- nomia reale e quella del sistema bancario. A questo scopo, forniamo una descrizione qualitativa e quantitativa delle dinamiche della leva nanziaria. Nel capitolo II, sfruttiamo un set originale di dati su 15 banche europee classi cate come G-SIB per valutare se l'espansione nei mercati esteri aumenta la loro rischiosit a, e attraverso quali canali si materializa. / The last global nancial crisis clearly illustrated the crucial role of interbank linkages in channel- ing and amplifying shocks hitting the system and, therefore, in the emergence of systemic risk. In this thesis, we present theoretical and empirical methodologies for analysing the potential for systemic risk in a interconnected banking network. The dissertation comprehends two essays on nancial networks and systemic risk and is organ- ised in two chapters. In chapter I, we analyse and model some complex interactions and feedback relationships within a nancial network, with the objective of delving into the linkages between fragility in the real economy and in the banking system. For this purpose, we provide a qualita- tive and quantitative description of leverage dynamics. In chapter II, we exploit an original dataset on 15 European banks classi ed as G-SIBs by the BIS to assess whether expansion in foreign markets increases their riskiness, and through which channels that eventually happens.

No description available.