تجزیه و تحلیل مبتنی بر گزینه های از انتظارات در حال ظهور در بازار نرخ ارز: طرح و مستغلات برزیل، 1994-1999

This paper uses currency option data to investigate market expectations on the Brazilian Real–U.S. dollar exchange rate from October 1994 through March 1999. We derive implied probability density functions (PDF) for expected future exchange rates and thus measures of the credibility of the “crawling peg” and target zone regimes governing the exchange rate. Target zone credibility was poor prior to February 1996, improved afterwards through September 1997 and later started to worsen again. The market anticipated periodic band adjustments and estimated distributions are very sensitive to political and economic news affecting the credibility of the regime.

This paper uses a new data set of options on the Brazilian Real/US dollar exchange rate to extract market expectations, as embodied in the risk-neutral probability density function (PDF), of Real–Dollar exchange rates over horizons of 1 to 3 months. Unlike ordinary exchange rate forecasts that provide only a point estimate of the future exchange rate, options-based forecasts, by permitting the derivation of a PDF, describe a range of realizations and the probability attributed to each range. This PDF-based approach is especially effective for an analysis of the Real/$ exchange rate, which from June 1994 to January 1999, under the Real Plan, was characterized by a combination of a crawling peg and a target zone regime. Over short horizons, the exchange rate followed a crawling peg surrounded by a “miniband”, but for long horizons, superimposed on the crawling peg, there was also an official “maxiband” with a fixed (non-crawling) central rate, floor, and ceiling. The PDFs derived in this paper enable us to compare market expectations embedded in options with these two concurrent regimes. For example, we can identify whether markets in fact anticipated a faster depreciation, and if so, where (relative to the crawling peg) probability was concentrated. Relative to a single point expectation of the future exchange rate, a great advantage of a full PDF is the ability to disentangle magnitude and probability of expected depreciation. For the longer-horizon fixed target zones, we conduct “arbitrage-based tests” of credibility, developed in Campa and Chang (1996), that are virtually assumption-free. Given these target zones, we are also able to determine both “intensities” and probabilities of realignment, and to investigate possible economic determinants of realignment intensity. Thus, a single approach using Dollar–Real options permits us to analyze both facets of the Brazilian exchange rate regime during the Real Plan. This work contributes to the growing literature on the use of options to characterize expected asset returns, and in particular to predict currency crises. Recent empirical work using options to identify the distribution of expected exchange rates includes Malz (1996b) and Campa et al., 1997 and Campa et al., 1998. Papers specifically focusing on currency crises, especially the 1992 ERM crisis, include Campa and Chang (1996), Malz (1996a) and Mizrach (1996). These can be contrasted against measures of devaluation risk not based on options, as in Bertola and Svennson (1993), Kaminsky et al. (1997) and Svensson (1991). The motivation for this research is twofold: first, to use options-based estimates of the PDF to compare market expectations with the two concurrent exchange rate regimes in the Brazilian Real Plan; and second, to observe the time path of market perceptions to gauge policy effectiveness over time. Furthermore, this is one of the first options-based tests of exchange rate regime credibility on an emerging market. Within emerging markets, this is also the first paper to deal with the data challenges of exchange-traded options, rather than over-the-counter (OTC) volatility quotes. OTC data, by construction free of arbitrage violations, are normally subject to less observation error and, hence, easier to interpret empirically. Thus, our technique could potentially be used for other emerging markets including those with only exchange-traded currency options. The remainder of the paper is structured as follows. Section 1 describes the theoretical background to deriving risk-neutral probability density functions (PDFs) from options and for target zone regimes, the derivation of realignment intensities and probabilities, as well as arbitrage-based tests of credibility. Section 2 discusses the Real Plan and pertinent historical background, including the “miniband” and “maxiband” regimes. Section 3 introduces our option data, provides summary statistics, and conducts a preliminary analysis. Section 4 investigates the behavior of the PDF over time, and in the context of a crawling peg, describes the probability and magnitude characterizing expected deviations from this regime. Section 5 addresses the “maxiband” target zones, estimated realignment intensities and probabilities, and arbitrage-based measures of credibility. Section 6 explores the empirical relation between estimated intensities and standard macroeconomic factors. Section 7 concludes.

This paper has used a new data set of exchange-traded options from March 1995 through January 1999 to derive risk-neutral probability density functions for the Real/Dollar exchange rate over horizons ranging from 1 to 3 months. The PDF is a superior indicator to a single point estimate of exchange rate expectations, such as a forward rate or survey-based forecast, in that it assigns varying amounts of probability to different possible outcomes. Although we introduce some approximations to compensate for sparse data, we make no assumptions about exchange rate dynamics. The PDF then can be used to analyze both the crawling peg and the maxiband exchange rate regimes. These two overlapping systems operated in Brazil since early 1995, several months after the June 1994 introduction of the Real Plan, designed to combat inflation and currency depreciation. In assessing market expectations under the crawling peg, we use the risk-neutral PDF to calculate both the intensity and probability of depreciation beyond the crawling peg. A high probability accompanied by a relatively low intensity, for example, indicates that the market anticipates depreciation beyond the peg, but most of this depreciation is concentrated just outside the peg. Empirically, we find that the credibility of the peg increased over time during 1996 and the first half of 1997 and started to deteriorate again after the start of the Asian crisis in July 1997. The occasional spikes in depreciation intensity and probability can usually be explained by identifiable political or economic news in Brazil. Our evaluation of the maxiband regime consists of two arbitrage-based tests of target zone credibility, as well as a measure of devaluation intensity outside the band. Tests based on arbitrage reject credibility whenever observed option prices are inconsistent with zero probability lying outside the band. When this occurs, devaluation intensity outside the band is positive. The numerical value of this intensity then provides a quantitative indicator of markets' questioning of the maxiband regime. Empirically, we are usually able to reject credibility, but find that through our sample ending in July 1997, the intensity of devaluation has fallen over time as the regime became increasingly credible. This paper also provides a more general methodology for extracting the risk-neutral PDF even when data are limited. In particular, we aggregate observations over several days, normalizing the option price by the contemporaneous forward rate. Our method involves fitting a single volatility smile to these multi-day observation periods. Assuming stationarity of the distribution over each period, this approach results in more precision when relatively few options are observed, a common difficulty with many emerging markets.10 Analysis of the shape of the PDFs over time also provides insight into market perceptions. In general, the PDFs appear to exhibit a greater degree of kurtosis and skewness (towards Real devaluation) with time. Increased kurtosis, i.e. fatter tails for a given level of volatility, suggests that increasingly markets believed that if a depreciation were to occur, it would be a large depreciation. Holding volatility constant, an increase in kurtosis implies less probability of a devaluation outside the target zone, but a larger expected devaluation if devaluation occurs. We also run regressions seeking to identify macroeconomic determinants of realignment risk. We find little evidence that standard macroeconomic indicators can explain observed realignment risk, consistent with Rose and Svensson (1994) and Campa and Chang (1998). Our observation of increasing kurtosis over time suggests that devaluation outside the band is increasingly perceived as a rare large event, rather than a more likely but not necessarily large event. Overall, the paper's findings reinforce earlier work on options' superior ability, relative to macroeconomic or interest-rate based indicators, to anticipate the periodic realignments of the exchange rate bands. By providing a more sensitive indicator of exchange rate risk—either in the form of depreciation beyond the crawling peg or a realignment of the maxibands—we have also documented the fluctuations in the degree of exchange rate credibility during the years of Brazil's Real Plan.