پیش بینی بحران ارز با استفاده از داده های بازار ADR: روش مبتنی گزینه محور

During capital control episodes, large price deviations between American Depositary Receipts (ADR) and their underlying stocks signal that a currency crisis is about to occur. We interpret this price spread as the price of a call option. Using option pricing theory we derive detailed information about both the probability of a currency crisis and the expected magnitude of devaluation. Analyzing daily ADR market data preceding the Venezuelan crisis (1996), our approach predicts crisis probabilities of almost 100% and forecasts the exchange rate after floating quite accurately. During the Argentine crisis (2002), the estimated exchange rates are similar to the actual ones.

Over the past few decades, many developing countries have fallen victim to financial crises. Due to their harmful effects, the literature has thus put an increasing amount of effort into developing models to predict the occurrence of such crises. This paper presents an approach for forecasting currency crises that relies on option pricing theory and uses high-frequency stock market data. In particular, we consider the price spreads between American Depositary Receipts (ADRs) and their underlying stocks in the emerging market. The most important branches of the literature on forecasting currency crises are based on either logit/probit models–which Eichengreen et al., 1995 and Eichengreen et al., 1996 and Frankel and Rose (1996) promote–or the signal or early warning indicator system put forward by Kaminsky, Lizondo, and Reinhart (1998) and Kaminsky and Reinhart (1999). Both types of model use observable economic and sociopolitical variables that presumably lead to currency crises. This body of literature not only provides useful insights into the nature and causes of currency crises, but also shows that currency crisis prediction in general is possible. Indeed, it appears that it would be possible to predict the occurrence of a currency crisis up to two years in advance with some accuracy. We focus on short term crisis prediction. The use of stock market data may yield noticeably more accurate forecasts than the models mentioned above, which are based on past observations of economic data describing the causes of a crisis. On the eve of a currency crisis, economic conditions change quickly and dramatically. The data used in existing models are generally observed infrequently and are often outdated. In contrast to these backward-looking approaches, the use of market data is forward-looking, since these data reflect market participants’ expectations. Moreover, the high frequency of stock market data more accurately represents changing conditions and makes daily crisis forecasts possible. The use of market data is based on the assumption that financial markets are able to efficiently process information about the value of the securities being traded, which implies that market prices should reflect all information available in this context, and that market participants are able to correctly anticipate exchange rate movements, at least on average. The market prices of securities whose value depends on the exchange rate can thus be used as input data to derive an assessment of currency crisis risk and the unobservable value of the exchange rate. Even if the markets themselves are not perfect in a theoretical sense, market data may still be superior to other sources of information. The ADR market in particular reveals information about exchange rate expectations, since the price difference between an ADR stock and its “original” stock is driven by the exchange rate. An ADR stock represents the ownership of a specific number of “underlying” or “original” shares in the home market on which the ADR stock is written.1 While the ADR stock is traded at a US stock exchange and is denominated in US dollars, the “original” stock is denominated in the currency and traded at the stock exchange of the home market. Every ADR stock can be converted into its respective “original” stock (and vice versa) through ADR conversion using custodian banks. Since the ADR stock and its corresponding “original” stock are substitutes through ADR conversion, and incorporate equivalent rights and dividend claims, both types of a company’s stock should exhibit the same price in exchange rate-adjusted terms. Previous to several currency crises, large and persistent price spreads developed between “original” stocks and their cross-listed ADR stocks. This happens when capital controls impede arbitrage forces and induce price segmentation. In this case, the law that an ADR stock and its “original” stock must have the same price no longer holds. Contemporaneously with the establishment of liquid ADR markets and the growing importance of ADR securities over the past decades, a comprehensive body of literature on ADR pricing has emerged. With respect to our topic, papers that are concerned with the relation between ADR prices and currency (crisis) risk are especially important. In analyzing daily data for a broad set of crisis episodes–including the United Kingdom (1992), Mexico (1994), Southeast Asia (1997), Russia (1998), and Brazil (1999)– Bin, Blenman, and Chen (2004), for example, find that the outbreak of a currency crisis in a given country leads to significantly negative abnormal returns on ADRs in that country. Bailey, Chan, and Chung (2000) analyze intraday ADR prices during the Mexican crisis of 1994/1995, and also report that ADR returns were (negatively) affected by news about the Mexican exchange rate regime and actual depreciations of the peso. Liang and Mougoue (2001) examine monthly data for 110 firms located in the UK, Japan, and South Africa from 1976 to 1990, and confirm that exchange rate fluctuations affect ADR prices. Several interesting studies, such as Arquette, Brown, and Burdekin (2008), Auguste, Dominguez, Kamil, and Tesar (2006), Levy Yeyati, Schmukler, and van Horen (2004) and Melvin (2003), consider the spreads between ADR and corresponding domestic stock prices, rather than the ADR prices themselves. They argue that these ADR spreads reflect the risk of a devaluation, since market participants can use the ADR market to hedge their funds (denominated in the domestic currency) against devaluation losses during periods in which capital control exists. Thus, they interpret the ADR spread as an indicator of devaluation risk, and discuss the development of this risk indicator during crisis periods. Some authors, such as Arquette et al. (2008) and Auguste et al. (2006), also analyze the determinants of ADR spreads. Our paper is inspired by these interesting contributions, but has a different focus. Instead of discussing the risk indication of ADR spreads during crisis periods and analyzing the underlying causes, we use the spreads to infer more detailed information about the crisis risk. We quantify crisis probabilities and the underlying value of the currency, which is unobservable because the exchange rate is pegged, by showing that an ADR portfolio–consisting of a short sold ADR stock and a (number of) corresponding domestic stock(s)–can be interpreted as either an option on the true value of the currency or a future on the actual currency value. Using the future approach, we estimate the actual value of the currency from observed values of the ADR portfolio. Based on the option approach, we derive the crisis probability on the one hand and the “true” value of the currency on the other. The use of ADR spreads for the estimation of the currency crisis risk and the true currency value using option pricing theory relies on the following considerations. To avoid a currency crisis, governments typically impose capital controls to impede cross-border capital flows. ADR conversion, however, provides a legal option to hedge proceeds–denominated in the emerging market’s currency–against devaluation losses during periods of capital controls. A call option position can be created by buying an emerging market (“original”) stock and selling short the corresponding ADR stock. For this call option position, the true exchange rate, or, more precisely, the value of the emerging market’s currency, represents the underlying instrument. Normally, if a currency is pegged, we do not know its “true” value. We can derive this value, however, using the option pricing formula. Then, we can use the estimated exchange rate series to derive the probability of devaluation, as well as the expected magnitude of the devaluation. Other input parameters for these calculations, such as the drift and volatility of the exchange rate, are derived simultaneously, where the estimated exchange rates are based on time series of ADR market data. To do this, we employ a maximum likelihood approach which relies purely on the model’s features and avoids additional assumptions.2 This options-based approach provides detailed information about the two dimensions of currency crisis risk: the probability that a currency crisis will occur and the expected magnitude of the devaluation, i.e. the “true” value of the currency. This information is critical for policymakers. If the probability of a currency crisis continues to increase, policymakers can implement the necessary measures to strengthen their currency and avoid a crisis. While the forecast approaches mentioned above serve a similar purpose, our approach additionally helps in the event that such actions fail or are not even undertaken. If the government prefers a pegged currency to a free float, it can realign the peg rate by an amount that our approach estimates as being appropriate for meeting market participants’ devaluation expectations. In doing so, the government can sustain the peg regime (with an adjusted peg exchange rate) and avoid a currency crisis. The turmoil caused by the Mexican crisis (1994/95), for example, could have been avoided. This crisis was triggered by a 15% devaluation of the peso, which was much lower than what market participants demanded, resulting in a speculative attack on the peso and eventually forcing the government to abandon the peg and allow the peso to float. Options-based approaches are widely used in other fields to forecast events influencing the value of financial claims related to different types of financial crises. When applied to banks and (the insurance of) their deposits, as Merton (1977) first proposed, option pricing theory can be used to forecast bank failures. Recent papers apply options-based (or structural) models to bank failures (see, e.g., Chan-Lau et al., 2004, Gropp et al., 2002 and Gunther et al., 2001) and to bank deposit insurance (see, e.g., Duan et al., 1995 and Duan and Simonato, 2002). Options-based models are also used in debt crisis forecasting and in estimating country default risk. Early examples are Clark (1991) and Claessens and van Wijnbergen (1993). These authors demonstrate that options-based approaches can be applied successfully by using economic fundamentals rather than market data. Examples of papers that apply market data to estimating options-based models of debt-crisis risk are Claessens and Pennacchi (1996), Huschens, Karmann, Maltritz, and Vogl (2007) and Keswani (2000). An influential and comprehensive body of literature deals with deriving information about future exchange rates based on market-traded currency options. By using option prices, the market’s expectations of future exchange rates (Campa, Chang, & Reider, 1998) or the correlations between exchange rates (Campa & Chang, 1998) can be estimated. Some pioneering papers have used market-traded currency options to forecast the occurrence of devaluations in managed exchange rate regimes. Campa and Chang (1996) and Malz (1996) use data on currency options to assess the realignment risk prior to the crisis of the European Exchange Rate Mechanism (ERM) in 1992–93. The authors find that option prices implied that the target zones of the mark-lira band and the mark-pound band lost credibility well before the lira and the pound were devalued. Campa, Chang, and Refalo (2002) were the first to apply option pricing theory to evaluate realignment risk in an emerging market. Using data on market-traded real-dollar options, the authors find that the market anticipated the steady realignments of the real-dollar exchange rate during Brazil’s crawling peg regime (1994–99). These papers show that it is possible to evaluate the risk of devaluation using data on market-traded currency options. We present an options-based approach that relies on the prices of synthetic currency options derived from price spreads between ADRs and their underlying stocks, which can be applied if no market-traded currency options exist. In fact, for many pegged currencies no market-traded currency options are available–be it because the emerging country is too small or because the currency peg is so credible that no demand for currency options exists–and, thus, the approaches mentioned above cannot be applied. In the last decade, however, many emerging market companies have begun to issue ADRs. Using the price spread between ADRs and their underlying stocks during capital controls, we can calculate the prices of artificial currency options and, thus, apply the options-based framework to estimate the probability of a currency crisis and the expected magnitude of the devaluation. We demonstrate the applicability of our framework for two currency crisis episodes: Venezuela in 1996 and Argentina in 2002. The remainder of the paper is organized as follows: Section 2 outlines a theoretical approach for inferring the price of a synthetic call option from price differences between ADR and “original” stocks. Section 3 introduces an option pricing framework that will enable us to derive the devaluation probability and expected exchange rates using the call option series. It also describes the estimation approach. Section 4 applies the approach and presents our findings, and Section 5 concludes.

We present and apply a framework which enables us to derive detailed information on currency crisis risk from ADR stock market data. On the one hand, we derive the probability of a currency crisis. On the other, we forecast the expected magnitude of devaluation by estimating the true value of the currency. To do this, we adapt option pricing theory to the problem of currency crisis forecasting by interpreting price differences between ADR stocks and their “original” stocks as a call option which can be used to hedge proceeds against possible devaluation losses. We also derive current estimates of the exchange rate that will actually prevail when the capital controls are lifted by interpreting the ADR portfolio as a currency future. Based on ADR market data, we estimate the parameters required to determine the probability and magnitude of devaluation by using a maximum likelihood approach. This approach exploits the features of the option pricing model, and, unlike other estimation approaches, requires no additional assumptions. In particular, this framework enables us to estimate the true value of a country’s exchange rate — which is otherwise not known when the currency is pegged and foreign exchange market data are not available. Since capital controls are often imposed to maintain a currency peg, the true exchange rate is an unknown quantity. For an example where a currency peg and capital controls existed simultaneously and where ADR market data are available, we consider the situation in Venezuela in 1995/1996. We obtain devaluation probabilities of around 95% and almost 100%, respectively, for four and two months prior to the outbreak of the Venezuelan currency crisis on April 22, 1996. With respect to the magnitude of the devaluation, we find that the expected exchange rates are very close to the initial floating values. Whereas in the Venezuelan case (1996) the true value of the currency is unknown, during the Argentine capital control period (2002) it is known. Shortly after capital controls were imposed, the Argentine peg collapsed. In this case, we find that the difference between the exchange rate value estimated using our framework and the actual market exchange rate is quite small, indicating reasonable results. Several robustness checks for our examples demonstrate that the results are relatively robust to moderate changes in the parameters. In the future, international firms and investors, as well as governments, could apply our approach to obtain a better assessment of devaluation risk and potential devaluation losses. Investors in international markets could detect profit opportunities from cross market arbitrage, which could emerge if other assets, like forward exchange rates or differentials between domestic and foreign interest rates, reflect a different expected exchange rate to the one derived from call option prices. Most importantly, however, our approach can help governments make the right decisions with respect to exchange rate policy. An increasing likelihood and estimated magnitude of a devaluation could be a wake-up call for policymakers to strengthen the value of their currencies. If the government is not able or willing to do this–and if letting the currency float is undesirable–it can realign the peg rate. In this case, the government can use our approach to identify the magnitude of devaluation the market demands, and in so doing sustain the peg arrangement and avoid a currency crisis.