Thiago Christiano Silva

• Ph.D.
• Senior Researcher at Banco Central do Brasil

Traditional supervised data classification considers only physical features (e.g., distance or similarity) of the input data. Here, this type of learning is called low level classification. On the other hand, the human (animal) brain performs both low and high orders of learning and it has facility in identifying patterns according to the semantic...

This book presents the features and advantages offered by complex networks in the machine learning domain. In the first part, an overview on complex networks and network-based machine learning is presented, offering necessary background material. In the second part, we describe in details some specific techniques based on complex networks for super...

In this paper, we propose novel risk-related network measurements to identify the roles that financial institutions play as potential targets or sources of contagion. We derive theoretical properties and provide a clear systemic risk interpretation for the proposed measures. Devised upon the notion of communicability in networks, we introduce the i...

We develop an innovative framework to estimate systemic risk that accounts for feedback effects between the real and financial sectors. We model the feedback effects through successive deterioration of borrowers’ creditworthiness and illiquidity spreading, thus giving rise to a micro-level financial accelerator between firms and banks. We demonstra...

We study how financing non-traditional local activities, conceived here as a proxy for activity diversification, is associated with economic growth. We use municipality-level data from Brazil, a country with large geographical, social, and economic disparities observed across its more than 5500 municipalities. We find that finance to non-traditiona...

Purpose In this article, the research objective is to empirically investigate the effect of the adoption of the Brazilian instant payment system, Pix, on the local credit market structure and the diversification of the banking system in Brazilian municipalities. Design/methodology/approach By analyzing the data, in this study, we compile and align...

Health literacy is a growing research area with specific aspects and different instruments to measure health literacy. This article uses natural language processing model to analyze the academic corpora regarding seven health literacy instruments - Health Literacy Questionnaire, Mental Health Literacy Scale, Rapid Estimate of Adult Literacy in Medi...

Detecting keywords in texts is a task of paramount importance for many text mining applications. Graph-based techniques have been commonly used to automatically find the key concepts in texts. However, the integration of valuable information provided by embeddings to enrich the graph structure has not been widely used. In this context, this paper a...

We study a large-scale Brazilian loan subsidy program to expand long-term credit. The government subsidizes banks’ funding costs for lenders, who then allocate credit to firms at regulated interest rates below a maximum ceiling. We propose and test a mechanism allowing banks to circumvent the rate caps and capture part of the subsidy. We show that...

The post-World War II decades experienced rapid growth in international trade, but a trend of weakening globalization has been consolidating recently. We construct the international trade network (ITN) using bilateral trade (2010–2022) to assess how interconnectedness has evolved in the face of recent developments. Our analysis reveals that, while...

Resumo: Este artigo avalia se a gestão municipal no Brasil do imposto incidente sobre terras rurais, o ITR, é afetada por ser o prefeito um ruralista. A hipótese de clientelismo indicaria que o prefeito ruralista tenderia a favorecer seus pares ao manter o estado de elevada sonegação e baixa arrecadação. Com dados municipais para o período de 2009...

The identification of key concepts within unstructured data is of paramount importance in practical applications. Despite the abundance of proposed methods for extracting primary topics, only a few works investigated the influence of text length on the performance of keyword extraction (KE) methods. Specifically, many studies lean on abstracts and...

In recent decades, health literacy has garnered increasing attention alongside a variety of public health topics. This study aims to explore trends in this area through a bibliometric analysis. A Random Forest Model was utilized to identify keywords and other metadata that predict average citations in the field. To supplement this machine learning...

In this study, we investigate the COVID-19 epidemics in Brazilian cities, using early-time approximations of the SIR model in networks and combining the VAR (vector autoregressive) model with machine learning techniques. Different from other works, the underlying network was constructed by inputting real-world data on local COVID-19 cases reported...

Introduction: In recent decades, health literacy, in connection with a broad range of public 1 health terms, has become a burgeoning field. This study aims to explore trends and biases in this 2 area through a bibliometric analysis. Methods: A Random Forest Model was utilized to identify 3 keywords and other metadata that predict annual citations i...

Access to clean water is crucial for human well-being and economic prosperity, while contaminated water has detrimental effects on various aspects of society. This paper examines the economic impacts of a catastrophic environmental disaster in Brazil-the collapse of the Mariana mining tailing dam. Using comprehensive microdata, including firm-to-fi...

The COVID-19 pandemic forced universities and schools worldwide to switch from in-person to online learning. This article analyzes the effects of the COVID-19 pandemic on education, particularly on the academic performance of undergraduate students in Brazil. The sudden shift from face-to-face to online learning, caused by government measures to co...

Oil markets reveal considerably volatile behaviour due to a range of factors. Exogenous factors, such as the COVID-19 pandemic and ongoing wars and conflicts, impose even more difficulties for prediction purposes. As a tool to better understand and improve forecasting models, many researchers are using sentiment analysis techniques to identify the...

Nestedness is one of the most pervasive and studied patterns observed in complex networks and refers to a hierarchical organization of the network. In this paper, we assess the determinants of the individual nestedness contribution (INC) in financial systems. To perform this task, we rely on data from two Brazilian financial networks: the bank-firm...

The identification of the most significant concepts in unstructured data is of critical importance in various practical applications. Despite the large number of methods that have been put forth to extract the main topics of texts, a limited number of studies have probed the impact of the text length on the performance of keyword extraction (KE) me...

This paper examines how government alerts about a potential dam rupture affect Brazilian municipalities’ local economic and financial conditions. The dam collapses in Mariana (2015) and Brumadinho (2019) revealed the extent of losses that environmental disasters can cause in terms of human lives, flora, and fauna. This paper investigates the effect...

O Ministério da Saúde (MS) brasileiro é responsável pelo Sistema Único de Saúde que produz continuamente uma grande quantidade de informações. No entanto, a descoberta de insights de seus bancos de dados ainda é incipiente. O uso de técnicas de Machine Learning possibilita explorar a capacidade de modelos preditivos para produzir conhecimento a par...

This study examines the relation between the COVID‐19 pandemic and hedge efficiency in commodities futures markets. In particular, we first evaluate the informational content of commodity futures by investigating whether futures prices are accurate and unbiased predictors of future–spot prices, and then we identify key financial and real economy tr...

Fiscal efficiency is determined not only by the municipality’s local characteristics but also by externalities from the neighborhood. In this paper, we study municipalities’ fiscal efficiency and assess the importance of such externalities in shaping local fiscal efficiency. The quantification of externalities is challenging because municipalities...

We investigate the impact of non-critical business and telework propensity on stock prices during the COVID-19 pandemic using panel data comprising 15,238 firms across 46 countries. After eight months of the COVID-19 outbreak, we find that firms operating in non-critical industrial groups have stock prices 6.52% lower than firms in the same subsect...

Detecting keywords in texts is important for many text mining applications. Graph-based methods have been commonly used to automatically find the key concepts in texts, however, relevant information provided by embeddings has not been widely used to enrich the graph structure. Here we modeled texts co-occurrence networks, where nodes are words and...

This paper investigates how the structure of the air transportation network affects air ticket prices in Brazil. Airlines serve specific airports in the vast Brazilian territory and, consequently, have unique air transportation network structures. Using more than 40 million airline-specific commercial flights, we extract topological patterns from t...

We explore a discontinuity in the incentives created by the Brazilian Tourism Regionalization Program. We employ a research discontinuity design, focussing on municipalities in the transitions between two classes. We observe significant effects of the Program in tourist cities, including an increase in the value-added of services and per capita GDP...

This paper examines churn prediction of customers in the banking sector using a unique customer-level dataset from a large Brazilian bank. Our main contribution is in exploring this rich dataset, which contains prior client behavior traits that enable us to document new insights into the main determinants predicting future client churn. We conduct...

The bursting of the US housing bubble in the second half of 2008 triggered an almost unprecedented systemic crisis in the world economy. The financial collapse quickly overflowed into the real economy and caused, among other effects, a sharp fall in the flow of world trade. Using export data from Brazilian municipalities, we show that the subprime...

We contribute to the literature on financial networks by presenting empirical evidence that the global shock of the COVID-19 pandemic caused changes in the forms and intensity of banking sector connections between different countries. These changes include providing the highest level of connectivity observed in the timeline initiated in 2005. We us...

The purpose of this paper is to assess the role of financial variables and network topology as determinants of systemic risk (SR). The SR, for different levels of the initial shock, is computed for institutions in the Brazilian interbank market by applying the differential DebtRank methodology. The financial institution(FI)-specific determinants of...

We study a novel economic network (supply chain) comprised of wire transfers (electronic payment transactions) among the universe of firms in Brazil (6.2 million firms). We construct a directed and weighted network in which vertices represent cities and edges connote pairwise economic dependence between cities. Cities (vertices) represent the colle...

Identifying time series patterns is of great importance for many real-world problems in a variety of scientific fields. Here, we present a method to identify time series patterns in multiscale levels based on the hierarchical community representation in a complex network. The construction method transforms the time series into a network according t...

We examine the high-frequency return and volatility of major cryptocurrencies and reveal that spillovers among them exist. Our analysis shows that return and volatility clustering structures are distinct among different cryptocurrencies, suggesting that return and volatility might have different spillover patterns. Further investigation via minimal...

We run an experiment to test how consumers of banking services value stress tests performed by their banks. We query respondents about the extent to which they would be willing to trade profitability if banks conduct stress tests, maybe for greater bank financial stability. Our paper connects and innovates in the banking literature by providing emp...

The interbank financial networks literature has been gaining ground since the 2007–2008 global financial crisis. This paper contributes to the literature of interbank financial networks by summarizing its trends and patterns of published scientific papers using a bibliometric complex network approach. We also provide a citation likelihood analysis...

We contribute to the finance-growth nexus literature by showing that credit origin, bank ownership, type of credit, and bank type matter in economic growth. We use a unique dataset covering 5,555 cities in Brazil, with granular information on credit characteristics. We find that non-earmarked credit to the corporate sector is associated with munici...

We open the black box of the monetary policy transmission mechanism with a granular model that considers the balance-sheet composition and network relationships of each economic agent. Though there are several well-documented channels through which monetary policy operates, we focus on the overlooked trading book channel, which arises because of ad...

We use a controlled experiment to analyze the impact of watching different types of educational traffic campaign videos on overconfidence of undergraduate university students in Brazil. The videos have the same underlying traffic educational content but differ in the form of exhibition. We find that videos with shocking content (Australian school)...

The impact factor has been extensively used in the last years to assess journals visibility and prestige. While it is useful to compare journals, the specificities of subfields visibility in journals are overlooked whenever visibility is measured only at the journal level. In this paper, we analyze the subfields visibility in a subset of over 450,0...

We examine the high-frequency return and volatility of major cryptocurrencies and reveal that spillovers among them exist. Our analysis shows that return and volatility clustering structures are distinct among different cryptocurrencies, suggesting that return and volatility might have different spillover patterns. Further investigation via minimal...

We propose a new methodology to evaluate the importance of fiscal risk to financial stability. We first develop a method to estimate the probability of non-compliance of public entities, which takes into account the strict legal framework that is mandatory for governments. While in our model the evolution of public entities' revenues is stochastic...

Although COVID-19 has spread almost all over the world, social isolation is still a controversial public health policy and governments of many countries still doubt its level of effectiveness. This situation can create deadlocks in places where there is a discrepancy among municipal, state and federal policies. The exponential increase of the numbe...

In this special issue, we welcome new insights, models,and applications in a wide variety of topics that bridge topicsin machine learning to complex economics and financenetworks. The application and adaptation of re-unsupervised learning methods, such as data and community clustering, ranking, anomaly detection, and semi-supervised and supervised...

Complex networks consist of a useful area for the development of research in the area of finance and economics. In economics and finance the focus is on the decision-making process for each agent and its consequences. These decisions can be analyzed from the point of view of networks, since they show interconnections between the decision-making pro...

We study a novel economic network comprised of wire transfers (electronic payment transactions) among the universe of firms in Brazil (6.2 million firms). We construct a directed and weighted network in which vertices represent cities and edges connote pairwise economic dependence between cities. Each city (vertex) represents the collection of all...

The impact factor has been extensively used in the last years to assess journals visibility and prestige. While the impact factor is useful to compare journals, the specificities of subfields visibility in journals are overlooked whenever visibility is measured only at the journal level. In this paper, we analyze the subfields visibility in a subse...

Resumo Neste artigo, apresentamos um estudo do impacto dos investimentos públicos sobre o crescimento econômico regional entre 2010 a 2016, com foco na região Meio-Norte do Brasil, mais precisamente, em 38 municípios do roteiro turístico Rota das Emoções, um produto de turismo integrado nos estados do Maranhão, Piauí e Ceará. A abordagem utilizada...

We model investor behavior by training machine learning techniques with financial data comprising more than 13,000 investors of a large bank in Brazil over 2016 to 2018. We take high-frequency data on every sell or buy operation of these investors on a daily basis, allowing us to fully track these investment decisions over time. We then analyze whe...

This special issue on financial networks seeks to bring novel methods and discussions to improve our understanding of financial markets. The bulk of the literature studies, interbank markets, or stock markets networks. This literature has provided important insights for the development of portfolio and risk management and financial regulation. Sinc...

We investigate the effectiveness of bail-in mechanisms in mitigating systemic risk and welfare costs to society during resolution processes. To perform this study, we define a network model of mutually exposed banks and use it to simulate the effects of shocks to these banks using granular data of the Brazilian banking system, its interbank exposur...

We analyze the impact of recessionary periods on Internet access of households in Brazil. We use microdata on Brazilian households and econometric techniques, such as pooled OLS and Probit regression models, to show that the Brazilian recession had a substantial negative impact on the Internet access of households. We find that, on average, the dem...

The present article analyzes how the Brazilian recession in 2014 affected internet access at home using annual data between 2012 and 2017 from a large scale, representative household survey by the Regional Center for Studies for the Development of the Information Society (Cetic). Pooled OLS and Probit regression models show that the Brazilian reces...

This study applies data envelopment analysis and the stochastic frontier approach to a sample of Indian commercial banks to discuss the inconsistencies between these models. We find that DEA average efficiency scores are, in general, lower than those from the SFA model. However, both models indicate similar trends on efficiency scores over the year...

In this work, we propose a methodology to decompose drivers of systemic risk that arise due to insolvency contagion in evolving financial networks. There is an ongoing discussion on how network topology and capital buffer influence systemic risk. On the one hand, the network contagion literature tends to emphasize the influence of the network topol...

Inflation targeting (IT) has recently been seen as one of the main causes of the authorities' unresponsiveness to the build up of financial imbalances during the recent financial crisis. We take data from banks from 66 countries for the period of 1998-2014 and compare how institutional quality as perceived by the national population impacts financi...

We simulate shocks to the real sector and evaluate how the financial system reacts and amplifies these events using unique Brazilian loan-level data between banks and banks and firms. Our analysis considers the feedback behavior that exists between the financial and real sectors through a micro-level financial accelerator. We find a strong "network...

Capturing financial network linkages and contagion in stress test models are important goals for banking supervisors and central banks responsible for micro- and macroprudential policy. However, granular data on financial networks is often lacking, and instead the networks must be reconstructed from partial data. In this paper, we conduct a horse r...

We conduct a horse race of methods to reconstruct financial networks from partial data. The various methods are assessed using financial network data obtained from 25 different markets, across 13 different jurisdictions. Our contribution is two-fold. We conduct a cross-country panel analysis of financial networks that sheds new and robust insights...

In this paper, we propose novel risk-related network measurements to identify the roles that financial institutions play as potential targets or sources of contagion. We derive theoretical properties and provide a clear systemic risk interpretation for the proposed measures. Devised upon the notion of communicability in networks, we introduce the i...

This study investigates to which extent results produced by a single frontier model are reliable, based on the application of data envelopment analysis and stochastic frontier approach to a sample of Chinese local banks. Our findings show they produce a consistent trend on global efficiency scores over the years. However, rank correlations indicate...

Our paper is the first to link bank liquidity performance and core-periphery network structures. We show that core-periphery structures can improve the liquidity performance of banks. We also find that network centrality plays a major role in bank liquidity performance. In particular, central players often display better liquidity performance than...