نوسانات نرخ ارز و تاثیر آن بر هزینه های معامله و نرخ بهره پوشش داده شده

This paper provides empirical evidence on the linkage between foreign exchange market volatility and daily 90-day covered interest rate parity (CIP) conditions of the three major exchange rates against the US dollar (US$). Markov regime shifting models were utilized to generate time series of volatility regime probabilities and these were used to explain the first and second moments of the daily deviations from and the transaction cost bands around the covered parity conditions. We find a significant positive relationship between the deviations and the regime probabilities, indicating an increasing probability of higher volatility state being associated with rising deviations (both first and second moments) from the parity condition. Similar positive relationship is found for the transaction bands. Rising (falling) probabilities of high (low) volatility regimes increased the first and second moments of the bands. Furthermore, we find a higher volatility state combined with a US$ depreciation is associated with significantly higher volatility in the daily deviations than an appreciation. Also, US$ depreciation is associated with widening transaction bands. This suggests that the level of market uncertainty was higher when the US$ was depreciating.

Short-term interest rate differentials between two countries, in theory, convey information regarding markets’ expectations concerning future exchange rate movements. This linkage between foreign exchange and money markets, via arbitrage, has been shown not to hold in its uncovered form (uncovered interest rate parity, UIP) but is often assumed to be a valid empirical regularity in its covered form (covered interest rate parity, CIP) once various market imperfections are taken into account.1 An enormous extant literature on forward premium (FP) (inter alia Hansen and Hodrick, 1980, Mark, 1988, Hodrick, 1989 and Backus et al., 1993; Levich and Thomas, 1993; Stulz, 1994, Bansal et al., 1995 and Bekaert, 1996) suggests that persistent deviations from the theoretical condition of UIP are deemed to be attributed to the failure of either or both of the two conditions underlying UIP, namely risk neutrality and rationality of expectations of market participants. However, despite some limited success in resolving the forward premium puzzle, on balance a general conclusion is that it remains a serious challenge and an anomaly in the literature of currency exchange. In parallel with the research on the forward risk premium briefly touched on above there exists a different strand of research that focuses on the empirical validity of the equilibrium conditions implied by CIP. Despite the empirical support in the Eurocurrency markets (Taylor, 1987 and Clinton, 1988), violations of CIP between national markets (significant deviations of forward exchange rates from the CIP conditions) are observed which are attributed to transactions costs (Frenkel and Levich, 1975, Frenkel and Levich, 1977, Frenkel and Levich, 1977 and Frenkel, 1981), political risks (Aliber, 1973), tax differentials (Levi, 1977), and capital market imperfections (Blenman, 1991). In addition, Balke and Wohar (1998) show non-linear dynamics of persistence of CIP deviations where deviations outside of the transaction cost bands were less persistent compared to those within the bands. While most of the arbitrage profits are small there are also large deviations from CIP from their sample. Peel and Taylor (2002) investigate covered interest arbitrage in the interwar foreign exchange market, using weekly US dollar (US$)–UKP rates during the 1920s. Specifically their analysis supports the Keynes–Einzig conjecture that the neutral band is of the order of one percent point on an annual basis and the deviations are moderately persistent even outside this band. Overall deviations from the exact CIP conditions are often attributed to transactions costs in line with much of the published work (see, inter alia, Demsetz, 1968, Frenkel and Levich, 1977, Frenkel and Levich, 1977, Taylor, 1987 and Taylor, 1989). In particular, in the Eurocurrency markets, where most of the market imperfections that hinder the covered interest arbitrage between national money/foreign exchange markets identified above are absent, deviations from the exact CIP are mostly attributable to transaction costs. In other words, no arbitrage bands around the exact CIP conditions are determined by transaction costs of covered arbitrage. The transaction costs, measured as bid–ask spreads, can also be influenced by risk considerations in the foreign exchange market. For example, an expected rise in the level of foreign exchange volatility would induce wider spreads to compensate the liquidity providers for the information (adverse selection) costs. Assuming that in the highly competitive Eurocurrency markets (in major currencies) a significant part of the transaction costs of covered arbitrage (and thus deviations from the exact CIP conditions) may then be attributable to this compensation for the information costs. In this paper, we aim to establish a time varying relationship between the foreign exchange market volatility and the transaction cost considerations in the covered interest rate parity conditions. We measure transaction costs in two ways, (i) deviations from the exact CIP conditions and (ii) vertical distance between the upper and lower bounds around the exact CIP conditions. The underlying motivation for using the first measure of transaction costs is the premise that rising levels of foreign exchange volatility due to market turbulence would give rise to wider bid–ask spreads of market makers to compensate for the information costs, and this would result in higher incidences of daily deviations. In addition, there is a potential for asymmetry of volatility impact depending on whether the volatility is associated with domestic/foreign currency depreciation/appreciation. To this end we examine the empirical relationship between the daily deviations from the exact CIP conditions and the nature of the foreign exchange market volatilities. In particular, we investigate disaggregate effects of various levels of spot exchange rate volatility (low, mid, and high levels) and the high and low volatility associated with US$ appreciation/depreciation on the daily deviations from the CIP conditions. This is accomplished by constructing various volatility states as required and investigating the explanatory powers of the generated (times series of) state probabilities from Markov regime switching models. In addition, we examine the time varying nature of the transaction cost bands around the CIP conditions. Transaction costs will vary across time depending on the level and the nature of volatility in each of the component markets. We investigate the empirical relationship between this time varying nature of the transaction bands and the time varying foreign exchange market volatility states. The major findings of this paper are that (i) there is a significant positive relationship between the daily CIP deviations and the exchange rate volatility regime probabilities in all three exchange rates we considered (US$ against the German Mark, the Pound Sterling, and the Yen) which indicates that an increasing probability of higher volatility (and decreasing probability of lower volatility) state is associated with rising deviations in both the first and the second moments, (ii) there is evidence that the combination of a higher volatility and a US$ depreciation is associated with significantly higher volatilities of deviations from the daily CIP than when the US$ was appreciating, (iii) in general, high volatility regime probabilities are associated with widening and more volatile transaction bands, and vice versa, and (iv) a US$ appreciation (especially with lower volatility regime) resulted in a narrowing bands whereas a US$ depreciation (and high volatility regime) significantly raised the first and second moments of the bands. The rest of the paper is organized as follows: in Section 2, data and modeling issues are discussed. We employ EGARCH models to address the statistical characteristics of the deviations from and the transaction cost bands around the exact CIP conditions, and the Markov regime shifting models are estimated to generate various volatility state probabilities in the foreign exchange markets which we use to help explain the first and second moments of daily deviations. Section 3 discusses the empirical results and the conclusions are presented in Section 4.

Empirical linkages between exchange rate volatilities and the time varying nature of daily covered interest rate parity conditions have been investigated in this paper. The time series of different volatility regime probabilities were used as explanatory variables to explain the first and second moments of daily deviations from CIP conditions and the transaction cost bands around the CIP conditions for the three currencies, the DEM, the YEN, and the UKP against the US$. We find significant explanatory power of these regime probabilities and the results of the estimations indicate that the higher is the probability of the foreign exchange market belonging to a higher volatility state, the higher is the conditional volatility of daily CIP deviations. We conjecture that this positive linkage is due to the heightened need to price in the extra risk in the forward contracts (in the form of a wider bid–ask spread) to mark a higher volatility state of the market. This view is consistent with the evidence of widening transaction cost bands in response to rising (falling) probability of high (low) volatility regime. It was also noticed that the combination of higher volatility and the US$ depreciation led to the highest rise in the conditional volatility of the deviations. Also, widening transaction band was associated with US$ depreciation and high volatility. This suggests a US$ depreciation is associated with higher level of uncertainty in the foreign exchange market necessitating higher spreads for market makers, and this could explain the widening and more volatile transaction bands as well as higher volatility of CIP deviations. Our contribution to the literature is the empirical confirmation of the hypothesis that there exists a positive relationship between spot foreign exchange market volatility and the risk premium/transaction cost considerations in the market from a research angle that has not been implemented previously. In particular, the volatility regime probabilities were shown to be an important factor in explaining the daily CIP deviations and the time varying nature of the transaction cost bands around the CIP conditions.