Data

To analyze the behavior of the cross-market premium, we start from a representative sample of emerging economies around the world that offer stocks with a long history of DR listings. These are Argentina, Brazil, Chile, Indonesia, Korea, Mexico, Russia, South Africa, and Venezuela. Most of these economies experienced the introduction (or lifting) of capital controls during the sample period.

We restrict our attention to stocks that are publicly traded both domestically and either on the NASDAQ or the New York Stock Exchange (NYSE) (the so-called level 2 and level 3 ADRs). From this set, we exclude the following stocks: (i) stocks that have less than two years of data (to impose some minimum data requirement), (ii) stocks for which the DR or the underlying security never trades (that is, we exclude stocks that always trade just on the domestic or the New York market, for which we would not be able to compute the crossmarket premium), and (iii) stocks that present irregular patterns in the time series (like stocks that display large unexplained shifts in trading volume). Aside from these criteria, we impose a minimum number of observations to estimate reliably the AR and TAR models; namely, stocks that have at least 100 and 500 observations for the AR and TAR estimates, respectively. This selection process leaves us with 98 stocks to compute AR estimates and 78 stocks to compute also TAR estimates. We collect data since 1990. For all countries, we have data up to 2004. However, for the case of Argentina, we also use data up to 2007 (in a separate section), to take into account the changes in the capital controls regime.¹⁵

For all our results, we use observations that exhibit contemporaneous trading, that is, observations corresponding to dates when trading takes place in both markets (the domestic market and New York). The decision to exclude other observations is a critical one given that many DRs from emerging economies display infrequent trading. While trading frequency per se should not be a concern for our purposes, adding non-contemporaneous trading can substantially alter our results (an issue that has been typically overlooked by the literature). The inclusion of observations with no trading in one of the markets may create variations in the cross-market premium that are entirely due to the fact that, in the absence of trading, the last traded price is repeated for non-trading days. In those cases, price disparities would reflect non-trading activity prices (specifically, valuations corresponding to different points in time) rather than differential valuation at the same time (the concept underlying the definition of the cross-market premium).¹⁶

In principle, one could argue that, for the non-contemporaneous trading observations, the premium is not arbitraged away because it belongs into the no-arbitrage band. If so, these observations would provide information about the band and should therefore be included in all our estimations; failing to do so would tend to understate transaction costs. However, the last traded price is generally not the contemporaneous (bid or ask) quote that a potential arbitrageur could actually trade on: if he/she were to profit from the price difference by buying in one market and selling in the other, he/she would find a different (probably narrower) cross-market premium. It follows that the inclusion of days with trading in only one market would tend to overstate transaction costs, the more so the higher the prevalence of these observations.

In light of this two-sided risk, we deliberately adopt a conservative approach and choose to include only contemporaneous trading days in most estimations. However, for completeness, we use all days (not jut the contemporaneous trading days) to compute the AR results in the first part of the paper, and show that our findings hold.

The data needed to calculate the premium (the dollar price of the stock in the domestic market, the price of the DR in New York, and the number of underlying shares per unit of the depository receipt) come from Bloomberg. For Argentina, Brazil, Chile, and Venezuela we use the closing price both in the domestic market and in New York. For Asian markets, which are already closed when New York opens, as well as for Russia and South Africa, we use instead the closing price (and the exchange rate) in the domestic market and the opening price in New York, to keep distortions due to time differences to a minimum.¹⁷

Download this Discussion Paper [ PDF 755.8KB| 39 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.