Investigating the nature of nonlinearity in Indian Exchange Traded Funds (ETFs)

Abstract

Purpose – The purpose of this paper is to first, test for nonlinearity in Local Indian Exchange Traded Funds (ETFs) listed at NSE, India – NIFTYBEES, JUNIORBEES, BANKBEES, PSUBANKBEES, and INFRABEES – using a battery of nonlinearity tests; second, to ascertain, using both metric and topological approaches, the adequacy of appropriate AR-GARCH models when it comes to capturing all of the nonlinearity in Indian ETFs; and third, to test for chaos in Indian ETFs. Design/methodology/approach – To start with, a battery of tests such as and limited to McLeod Li test, Engle's LM test, Tsay F-test, Hinich Bispectrum Test and Hinich Bicorrelation test were employed to test for nonlinearity in Indian ETFs. Subsequently, the nature of nonlinearity in all the ETFs was systematically investigated by subjecting the ETF data sets to a metric (BDS test) and a topological test (close returns tests) at different stages of the model-building process. Finally, Lyapunov Exponent test was employed to test for chaos in Indian ETFs. Findings – Test outcomes pertaining to a battery of nonlinearity tests indicate prevalence of nonlinearity amidst all ETFs except for INFRABEES. BDS test outcomes at the different stages of the model-building process indicated high sensitivity of the test outcomes to choice of embedding dimension, threshold value and residual transformations. Close returns test outcomes indicated that, but for BANKBEES, all of the nonlinearity in Indian ETFs could be captured by appropriate GARCH models. Finally, chaos was found to be absent in any of the ETFs considered for this study. Practical implications – The collective take-way from this study is threefold in nature. First, in light of the many limitations of the BDS test, topological approaches such as close-returns test offer a better avenue to test for adequacy of AR-GARCH models in explaining the nature of nonlinearity in asset price movements. Second, adequacy of AR-GARCH models in capturing all of the nonlinearity in NIFTYBEES, JUNIORBEES, PSUBANKBEES, and INFRABEES, as indicated by close-returns test findings, is a reflection of multiplicative nature of nonlinearity in these five ETFs. Third, persistence of nonlinearity in AR-GARCH filtered standardized residuals of BANKBEES, coupled with the absence of chaos in any of the ETFs considered for this study, brings to light the possibility of existence of additive nonlinearity in conjunction with multiplicative nonlinearity. Originality/value – This is possibly the first study that systematically investigates the nature of nonlinearity in Indian ETFs and ascertains the adequacy of AR-GARCH models when it comes to capturing all of the nonlinearity in Indian ETFs using a topological approach.

A study on volatility and return spillover of exchange-traded funds and their benchmark indices in India

“…Broad-based indices, as well as the sectoral indices-based index returns, are negatively influenced. However, the study of Madhavan (2014) shows that no asymmetric effect and ARCH effect are present in the BANKBEES and JUNIORBEES ETF. However, in contrast, our study finds both ARCH and asymmetric effect in BANKBEES and JUNIORBEES.…”

Nonlinearity in Global Crude Oil Benchmarks: Disentangling the Effect of Time Aggregation

“…Following the precedence in literature (Aghababa & Barnett, 2016; Lim, 2009; Lim & Brooks, 2009; Madhavan, 2014; Varghese & Madhavan, 2019), we employ the nonlinearity toolkit proposed by Patterson and Ashley (2000). To be specific, we employ a battery of nonlinearity tests, namely BDS test, Engle’s lagrange multiplier test, Hinich bicorrelation test, Hinich bispectrum test, McLeod–Li test, and Tsay test, so as to test for nonlinear serial dependencies in Brent, Dubai and WTI crude oil markets.…”

“…Gershenfeld and Weigend (1993) have found in their experimental analysis that, due, in part, to their inherent inability to capture the hidden nonlinear patterns in price series, studies that use the conventional econometric and statistical procedures are bound to demonstrate poor prediction performance and thereby fail to produce reliable forecasting results (also refer Mantegna & Stanley, 1999). Subsequent studies in the recent years have extensively studied, documented and have firmly established the price dynamics in most financial variables to exhibit a rather complex behavior, deviating significantly from that of a pure geometric Brownian motion (Hsieh, 1991; Lim & Hooy, 2013; Madhavan, 2014; Varghese & Madhavan, 2020). However, similar studies in commodity markets have been, at best, weak.…”

Long Term Dynamics of Indian ADRs Market: The Case of Persistence and Irregular Cycles

“…The stock market crash of 1987 provided further impetus to use of non-linear methods in financial research aimed at testing the validity of the random walk hypothesis (Lima 1998). In recent years, there has been spurt in studies which have refuted random walk and have documented nonlinear dynamics in a variety of financial return series (Ozer and Ertokatli, 2010;Mishra et al, 2011;Webel, 2012;Yilanci, 2012;Lim and Hooy, 2013;Madhavan 2014). However, existing literature provides mixed evidences on existence of non-linearity in financial return series.…”

“…The incomprehensiveness of literature in the area of persistence lies in the fact that the major focus of these studies is equity markets, and very few focus on other types of security viz. bonds, FX or DR markets (Malkei, 2003;Beine & Laurent, 2003;Oh, Kim & Eom, 2006;Kumar and Maheswaran, 2013;Madhavan, 2014;Anagnostidis and Emmanouilides, 2015;Ferreira & Dionisio, 2016;Sensoy & Tabak, 2016;and Masa and Diaz, 2017).…”