TeX 62.4%
Python 37.5%
1% ============================================================================2\section{Literature Review}3\label{sec:literature}4% ============================================================================56\subsection{Volatility Spillovers and Connectedness}78The study of volatility transmission across financial markets has a long intellectual history, beginning with the recognition that volatility clusters \citep{Mandelbrot1963,Fama1965} and exhibits contagion effects across markets \citep{King1990,Forbes2002}. Early approaches relied on multivariate GARCH models \citep{Engle2002,Baele2005} and regime-switching frameworks \citep{Ang2002} to capture cross-market volatility dynamics. While these models provided important insights, they were limited in their ability to decompose directional spillovers and identify which markets serve as net transmitters versus receivers.910The introduction of the \citet{DieboldYilmaz2009} spillover index marked a methodological breakthrough. Building on the concept of forecast error variance decompositions from VAR models, they proposed a single summary measure---the total spillover index---that captures the proportion of total forecast error variance attributable to cross-market shocks. \citet{DieboldYilmaz2012} refined this framework by adopting the generalized VAR approach of \citet{Koop1996} and \citet{Pesaran1998}, which eliminates the sensitivity to variable ordering inherent in Cholesky decompositions. This generalized framework yields an asymmetric connectedness matrix that provides directional ``TO'' and ``FROM'' spillover measures for each variable, as well as net directional spillovers.1112\citet{DieboldYilmaz2014} further extended the methodology to enable network visualization of the connectedness structure, drawing on tools from graph theory. Their approach has since been applied extensively across financial markets: \citet{Demirer2018} examined global bank stock return and volatility connectedness, \citet{Barunik2018} proposed a frequency-domain decomposition of connectedness to distinguish short-run from long-run spillovers, and \citet{Antonakakis2020} applied time-varying parameter (TVP) VAR models to capture smooth evolution of connectedness without the need for rolling windows.1314\subsection{Cross-Asset Volatility Transmission}1516A growing body of literature examines volatility transmission across different asset classes, motivated by the observation that cross-asset correlations tend to increase during market stress \citep{Longin2001,Ang2002}. \citet{Kang2017} investigated spillovers between commodity and equity markets, finding that oil and gold serve as significant volatility transmitters during crisis periods. \citet{Diebold2012} studied connectedness among four major asset classes and found that return spillovers are more important than volatility spillovers for most market pairs.1718The foreign exchange market, as the world's largest and most liquid financial market, has been shown to play a central role in cross-asset volatility transmission. \citet{Greenwood2014} documented that carry trade unwinds can generate volatility spillovers from FX to equity markets, while \citet{Barunik2016} found significant volatility transmission between the EUR/USD exchange rate and European equity indices. The Japanese yen, as a traditional safe-haven currency, has been shown to exhibit distinctive volatility dynamics during crisis periods \citep{Ranaldo2010,Habib2012}.1920Commodity markets, particularly crude oil and gold, have been extensively studied as both shock transmitters and hedging instruments. \citet{Vivian2012} found significant time-varying volatility spillovers between oil and stock markets across multiple countries. \citet{Baur2010} established gold's role as a safe haven during extreme equity market downturns, while \citet{Reboredo2013} documented asymmetric spillovers between gold and oil prices.2122\subsection{Cryptocurrency and Traditional Market Integration}2324The emergence of cryptocurrency markets has introduced a new dimension to cross-asset volatility analysis. Early studies found limited connectedness between Bitcoin and traditional financial assets \citep{Bouri2017,Corbet2018}, supporting the diversification benefits of including crypto in multi-asset portfolios. However, more recent evidence suggests increasing integration, particularly during market stress periods.2526\citet{Bouri2019} examined volatility spillovers between Bitcoin and a range of financial assets, finding that crypto--traditional market connectedness increased significantly after 2017. \citet{Corbet2020} analyzed the contagion dynamics between cryptocurrencies and conventional financial markets during the COVID-19 pandemic, documenting a sharp increase in crypto--equity correlations. \citet{Yousaf2022} extended this analysis to include a broader set of cryptocurrencies and found evidence of increasing volatility transmission from crypto to commodity markets.2728The question of whether crypto markets have transitioned from volatility receivers to originators has important implications for systemic risk monitoring. \citet{Ante2021} provided evidence of Bitcoin's growing influence on traditional markets, particularly through the channel of institutional adoption and the introduction of crypto-linked exchange-traded products. More recently, \citet{Makarov2020} documented the role of arbitrage and cross-exchange price discovery in the crypto market microstructure.2930\subsection{Intraday Volatility Dynamics and Trading Sessions}3132The use of high-frequency data to study volatility dynamics has a rich tradition in financial econometrics. \citet{Andersen2001,Andersen2003} pioneered the use of realized volatility constructed from intraday returns as a model-free measure of ex post volatility. The optimal sampling frequency for realized volatility estimation has been debated extensively, with \citet{BarndorffNielsen2002} and \citet{Zhang2005} addressing the trade-off between discretization bias and microstructure noise.3334Intraday volatility patterns exhibit well-documented regularities across asset classes. Equity markets display a U-shaped or reverse J-shaped intraday volatility pattern, with elevated volatility at the open and close of trading \citep{Wood1985,Harris1986}. FX markets, trading 24 hours, show volatility peaks during the overlap of major trading sessions \citep{Andersen1998}. Cryptocurrency markets, as continuously traded assets, display distinctive intraday patterns that reflect their global, decentralized nature \citep{Eross2019}.3536The interaction between trading sessions has been studied primarily in the context of equity markets. \citet{Hamao1990} found evidence of volatility spillovers from the Tokyo Stock Exchange to London and New York, while \citet{Lin1994} documented bidirectional volatility transmission between the New York and Tokyo stock markets. More recently, \citet{Fry2022} examined how Bitcoin's 24/7 trading provides a continuous information channel that can bridge traditional market close-to-open gaps.3738\subsection{Contribution to the Literature}3940This paper contributes to the existing literature in several important ways. First, we provide the most comprehensive high-frequency multi-asset connectedness analysis to date, spanning four asset classes with 5-minute data over an 18-year period. Second, we document the time-varying evolution of the volatility transmitter--receiver hierarchy, with particular attention to the changing role of cryptocurrency markets. Third, we exploit the 24/7 nature of crypto trading to examine session-specific volatility spillovers, providing novel evidence on the role of trading hours in shaping cross-asset volatility dynamics. Fourth, we demonstrate that the Diebold--Yilmaz connectedness framework, applied at high frequency, can serve as an early warning indicator for systemic risk buildups.41