There is a vast literature on asset return predictability in a linear regression framework. However, there is increasing evidence that asset returns may be better characterized by a model which allows for nonlinear behavior. In this paper, an attempt is made to examine the nonlinear behavior of returns of four exchange rates of Indian rupee with respect to Euro, British Pound, US Dollar and Japanese Yen, by utilizing tests based on nonlinear modeling. The application of BDS test strongly rejects the null hypothesis of independent and identical distribution of the return series as well as the ARMA residual series for all the exchange rates under study. This is consistent with nonlinear dependence in the return series. However, it is not enough to merely identify nonlinear dependence. So, the study investigates whether the nonlinear dependence is caused by predictable conditional volatility. It is found that GARCH(p, q) model fits all the market return series adequately and accounts for the nonlinearity found in the series.

TIn the literature on asset return, many studies have employed traditional statistical tests such as autocorrelation tests, and tested the Efficient Market Hypothesis (EMH) mainly focusing on the linear predictability of future asset price changes. If the later turn out to be uncorrelated then the EMH is accepted and the asset market in question is deemed to be informationally efficient, and if they are found to be serially correlated, the EMH is rejected and the market is considered inefficient. However, researchers like Brock et al. (1991 and 1992) point out that lack of linear dependence does not rule out nonlinear dependence, which if present, would contradict the random walk model. Evidence of this possibility is provided by Granger and Andersen (1978) and Sakai and Tokumaru (1980), who demonstrate that nonlinear models may exhibit no serial correlation, while containing strong nonlinear dependence. In fact, as Campbell et al. (1997) argue many aspects of economic behavior may not be linear, and may cause rejection of independent and identical distribution (iid). There may be several reasons behind the nonlinear behavior of financial markets. First, market imperfections and some features of market microstructure may lead to delays of response to new information, implying nonlinearity in asset price changes. Schatzberg and Reiber (1992) suggest that asset prices do not always adjust instantaneously to new information. For instance, transaction costs may make investors unwilling to respond rapidly to the arrival of new information. In turn, they would rather wait until their expected excess profits (net of transaction cost) are high enough to allow for positive returns. This delay in adjustment may lead to nonlinearity in asset price changes. Further as Shleifer and Summers (1990) argue, there are two types of investors in the market: rational arbitrageurs or speculators, who trade on the basis of reliable information, and noise traders, who trade on the basis of imperfect information. Because of the informational asymmetries and lack of reliable information, noise traders may lean towards delaying their responses to new information in order to assess informed traders’ reaction, and then respond accordingly. Based on the above line of thinking, one may assume that economic systems may be nonlinear. The presence of nonlinear dependence may have short term, if not long term forecasting potential, provided the actual generating mechanism is known. This paper attempts to test whether the rupee exchange rates present nonlinear behavior and if so, what is its nature.

The paper is organized as follows: it presents a brief review of the related literature explaining the motive of the study. Subsequently, it discusses the data and methodology used in the paper, followed by a discussion of the findings, and finally, the conclusion is offered.

Applied Economics Journal, Nonlinear Dependence, Conditional Volatility, Indian Rupee Exchange Rate