اثبات در پودینگ: داوری در بازارهای نوظهور محدود ممکن است

We investigate whether arbitrage trades exist in emerging markets with trading barriers. Using two-year intraday data for 16 Argentinean and Egyptian depository receipts and their underlying stock, we find large intraday deviations from parity. We extend the standard arbitrage identification procedure to account for volumes and precise dynamic measures of trading costs, resulting in 9.81% and 15.32% of Argentinean and Egyptian matched trades identified as arbitrage opportunities, which we show, result in real profitable arbitrage trades. Arbitrage profits of USD 1.8 million from Argentinean and USD 1.2 million from Egyptian depository receipts were estimated over the sample period.

Arbitrage, the simultaneous purchasing and selling of identical assets to take advantage of price differences, has been referred to as “one of the central concepts of financial economics” (Mitchell et al., 2002). In this paper, we study arbitrage in the market for emerging market depository receipts (DRs) and examine its role in the efficient pricing of those cross-listed equities. In theory, since DRs and their underlying stock represent identical claims on a firm's cash flows, arbitrage should keep their prices in parity, as long as arbitrage costs are sufficiently small. In financial models, the costs of arbitrage are typically assumed to be zero, so arbitrage opportunities disappear almost as quickly as they appear. As prices diverge, arbitrageurs intervene to bring prices back to parity by buying the underpriced security in one market and selling the other at a higher price in the other market, thereby making risk-free profit. A unique feature of DRs that ensures efficient arbitrage is their fungibility, which allows arbitrageurs to exchange freely between the cross-listed pair. We examine two central questions involving the efficiency of arbitrage operations in the DR market. The first is whether arbitrage opportunities exist between DRs and their underlying stocks. The lack of such opportunities confirms that markets are efficient and integrated since prices of identical securities align across different markets. Should we find evidence of arbitrage opportunities, however, their existence only compromises market efficiency if such risk free profitable opportunities persist without being utilized or explained by microstructure limitations to arbitrage. This poses the second question on how arbitrage opportunities disappear and the role of arbitrageurs in this process. Most of the empirical literature on arbitrage so far has focused on studying the first question of whether arbitrage opportunities exist. Early studies rely on daily closing prices and find no evidence for significant deviations from parity between DRs and their underlying stocks (Rosenthal, 1983, Kato et al., 1991 and Park and Tavakkol, 1994). Yet because different markets have different trading hours, a daily analysis provides a more global picture since it usually compares prices that do not occur in the same point in time. This motivated more recent studies to rely on intraday data. They still find either zero arbitrage opportunities (Miller and Morey, 1996) or extremely small, infrequent, and short-lived ones (Suarez, 2005a). These previous results are not surprising given that those studies use samples from developed market DRs in which trading costs are relatively low, liquidity is relatively high, and trading barriers are absent. But these conditions do not hold in many emerging markets, and emerging market DRs have increasingly come to dominate foreign cross-listing (Global Finance, 2010). A recent study by Gagnon and Karolyi (2010) provides evidence for large deviations from parity in emerging market DRs. Such deviations were attributed to large transaction costs and trading barriers that could limit arbitrage operations from taking place. Their underlying data are sampled daily, however, and “as a result (they) cannot know for certain whether the patterns in price deviations are economically real or artifacts of asynchronous trading between the two securities”. Our paper contributes to the arbitrage literature as follows. First, it extends previous studies studying whether arbitrage opportunities exist by conducting the first intraday contemporaneous analysis on DRs from emerging markets. We provide evidence on the presence of large arbitrage opportunities, after accounting for various limits to arbitrage present in those markets, and confirm the hypothesis that they appear in markets exhibiting lower degrees of efficiency and integration (Suarez, 2005a). Second, our study provides an original empirical analysis on the role of arbitrageurs in eliminating arbitrage opportunities. This was so far a largely unexamined question that draws on a slim body of literature that tests whether trades are important for price convergence between cross-listed stocks (Kaul and Mehrotra, 2007). Our results show the importance of arbitrage trades in eliminating arbitrage opportunities. This study focuses on two emerging markets: Argentina and Egypt. We use two years of intraday data on Argentinean stocks listed on the Buenos Aires Stock Exchange (BCBA) and cross-listed as American DRs (ADRs) on US exchanges, as well as Egyptian stocks listed on the Egyptian Stock Exchange (EGX) and cross-listed as Global DRs (GDRs) on the London Stock Exchange (LSE). The inclusion of different host and destination countries enables us to compare results across settings. Furthermore, our sample overcomes asynchronousity problems by focusing on overlapping periods when arbitrageurs could trade both the DR and the underlying stock. In the emerging markets we study, both local markets allow full fungibility1 between the DR and its underlying stock, ensuring feasibility of arbitrage operations. Despite this, limits to arbitrage in those markets include high trading costs and short-sales restrictions (Bris et al., 2007). Argentina also has capital controls, although Egypt did not at the time of our study.2Gagnon and Karolyi (2010) argue that arbitrage will be difficult to conduct in the presence of restrictions to short selling in the home market when the DR is selling at a discount to the stock. Moreover, Yeyati et al. (2009) discuss the limiting effects of capital controls on arbitrage, since they prevent arbitrageurs from moving cash to the local market to buy the underpriced security. We therefore use our sample to provide the first empirical intraday study that analyzes whether those restrictions truly hamper arbitrage activities. In our sample, we find large and statistically significant intraday price deviations as high as 57% between Argentinean DRs and their underlying stock and 24% for Egyptian ones. Our markets are characterized by various limits to arbitrage, and therefore in answering the first question on whether arbitrage opportunities exist in our sample, we only identify a price deviation as an arbitrage opportunity if it exceeds a ‘no-arbitrage’ band, defined as a threshold that incorporates the large trading costs and illiquidity present in our markets. We extend the standard arbitrage identification methodology to account for those and find that approximately 9.81% of Argentinean and 15.32% of Egyptian price deviations can be classified as profitable arbitrage opportunities. These linger for an average of 14 min in Argentinean securities and 46 min in Egyptian securities, and it takes on average 1.9 trades in Argentinean DRs and 1.58 trades in Egyptian DRs for them to disappear. These statistics vary within the sample, with arbitrage opportunities in more liquid and active securities persisting for shorter periods. The evidence on the presence of large and frequent arbitrage opportunities in our sample allows us to examine the mechanism by which they disappear. We therefore examine the second question on whether such arbitrage opportunities involve real arbitrage trades that contribute to the convergence of prices to the no-arbitrage bands. Our methodology relies on the application of a filtering algorithm on our transaction data, which extracts real arbitrage trades (Fig. 2). They reveal the active role of arbitrageurs in aligning prices A conservative arbitrage trading strategy reveals profits of approximately USD 1.8 million and USD 1.2 million from arbitrage operations involving Argentinean and Egyptian DRs, respectively, over the two-year period we analyzed. This paper is organized as follows. Section 2 presents our data description and price deviation construction. In Section 3, we present our tests and the results of our arbitrage analysis; and Section 4 provides a discussion and the conclusions of our findings.

This paper contributes to pre-existing studies on arbitrage in the DR market in several ways. It supplements Gagnon and Karolyi's (2010) study by dissecting their observations on the presence of large price deviations in cross-listed emerging market equity with limits to arbitrage, to show that arbitrage opportunities do exist. It confirms the hypothesis set forth in Suarez (2005a): that cross-listed stock with lower trading frequencies presents larger arbitrage opportunities. Furthermore, it provides empirical support for Kaul and Merhotra's (2007) results on the role of trades in price convergence. To our knowledge, this is the first evidence of the importance of arbitrage trades in eliminating arbitrage opportunities in DRs. The novelty of our results is attributed to the use of intraday data as well as extending the arbitrage identification methodology to better fit the microstructure limitations present in our sample of emerging markets. We reach a point that necessitates a discussion of why large arbitrage opportunities appear in our sample. The first involves an argument of market efficiency. Complete market efficiency implies that riskless profitable arbitrage opportunities are non-existent due to the active participation of rational and information efficient market players, with the threat of arbitrage usually enough to keep prices aligned. While Suarez (2005a) finds small incidences of arbitrage opportunities in developed market securities they are not far from full efficiency. However, he expects that “all other cross-listed stock markets which are less developed and include less heavily traded stocks to exhibit lower degrees of efficiency and integration”. We extend his work and test his hypothesis. Indeed, previous studies on the efficiency of our sample of emerging markets show that there are significant deviations from efficiency and that these markets are at least weak form efficient (see studies by Mecagni and Sourial (1999) and Lagoarde-Segota and Luceya (2008) on the efficiency of the Egyptian Stock Exchange and Rossi (2000) on the efficiency of the Buenos Aires Stock Exchange). Therefore, one reflection of the low levels of market efficiency of our markets is the presence of the large frequent and profitable arbitrage opportunities we document in this work. Arbitrage opportunities between DRs and their underlying stocks should not appear in the first place if movements in those identical securities only reflect changes in fundamental value. Yet recent empirical evidence shows that differential co-movements between identical stocks trading in different markets can reflect behavioral or noise trading (Barberis et al., 2005) as well as differences in ownership base. In the case of identical securities trading in two different markets, arbitrage opportunities can appear due to the differential co-movements of the DRs and their underlying stock price, since as Gagnon and Karolyi (2010) argue, there are “little reasons to believe that the two groups of noise traders would share the same mistaken beliefs about the shares’ fundamental value”. Besides behavioral trading, another reason explaining the different co-movements between two DRs and their underlying stock are differences in the ownership base, since institutional and individual traders do not exhibit similar trading patterns (Ji, 2005). Finally, information asymmetry between investors in different markets can explain the presence of arbitrage opportunities. If an investor in the local market has information that will make the stock price soon drop, he might quote well below the current market price, creating an arbitrage opportunity (Suarez, 2005a). Information asymmetry between investors in the local versus the foreign market is examined with a price discovery analysis for whether the local or foreign market leads. Indeed, Ansotegui et al. (2013) show that for all cross-listed Egyptian equities and the majority of Argentinean ones, the local market leads in the sense that DR prices adjust more often to movements on the local stock than vice versa, reflecting the flow of information between markets. In summary, this paper establishes the presence of large profitable arbitrage opportunities in DRs from emerging markets using intraday data and show the importance of arbitrage trades in eliminating them. The evidence on the actual profitability of arbitrage trades in emerging markets was probably concealed so far to profit from such an anomaly. For individuals to take a piece of the pie may prove difficult, however, as this activity must be institutionalized by large traders with large capital holdings in order to overcome the existing limitations. Future studies can explore whether arbitrage activity persists in other samples of DRs around the world.