Endnotes

¹See Gozzi et al. (2006), Levine and Schmukler (2006 and 2007), and references therein.

²Note that the cross-market premium is not a measure of capital mobility. In a world of perfect capital mobility (i.e., with no restrictions to the cross-country movement of capital), effective integration (price convergence) would still be affected by the intensity of transaction costs.

³Errunza and Losq (1989) describe some other channels through which capital controls may affect asset prices. They argue that, from a global diversification perspective, capital controls impede investors to hold directly country-specific risk. This would affect the price of securities after controls are dismantled, due to the probable rebalancing of investors’ portfolios towards more efficient ones.

⁴Depositary receipts have been used recently to assess the impact of capital controls. Rabinovitch et al. (2003) attribute the persistence of return differentials between ADRs and stocks in Chile to the presence of controls. Melvin (2003) and Auguste et al. (2006) examine the large ADR discounts that built in the midst of the Argentine crisis in early 2002, which Levy Yeyati et al. (2004) interpret as a reflection of the strict controls on capital outflows and foreign exchange transactions imposed at the time. We explore this hypothesis in more depth here.

⁵The importance of non-linearities in the behavior of asset prices has received ample attention in the literature. For example, Froot and Obstfeld (1991) construct a model in which government intervention leads to nonlinearities in the pricing of the foreign exchange rate. Sercu et al. (1995) build a model with a no-arbitrage band for the nominal exchange rate around its purchasing power parity value.

⁶Although TAR models have mostly been used in the PPP literature, more recently the model has also been applied to financial data. Rabinovitch et al. (2003), for example, use a TAR model as an approximation for the arbitrage adjustment mechanism between the local and ADR markets for Argentine and Chilean stocks. Canjels et al. (2004) use a TAR model to study the efficiency of the dollar-sterling gold standard and provide insights into the evolution of market integration in the classical gold standard. In addition, several studies have applied TAR models to study the behavior of interest rates (Balke and Wohar, 1998, Mancuso et al., 2003, Juhl et al., 2006, among others).

⁷See Imbs et al. (2003) for single-good price comparisons.

⁸Imbs et al. (2005) argue that this bias explains the slow convergence to the purchasing power parity (PPP) literature.

⁹In addition to price-based measures, stock-based measures of financial integration have spawned a large body of empirical work. A thorough survey of the vast literature on measuring financial integration far exceeds the scope of this paper. A comprehensive overview of the main operational measures of financial integration is provided by Obstfeld and Taylor (2002) and Prasad et al. (2003), among others.

¹⁰Studies based on stock market indexes include, among many others, Cashin et al. (1995), Soydemir (2000), Masih and Masih (2001), Scheicher (2001), and Chen et al. (2002). Capital asset-pricing models to test for market integration have been applied by Bekaert and Harvey (1995), Bekaert et al. (2005), and Carrieri et al. (2007), among others.

¹¹Criteria such as the (covered and uncovered) interest rate parity, and the real interest rate parity conditions, are related to this group to the extent that they focus on the analysis of onshore-offshore return differentials (see, among many others, Meese and Rogoff, 1988, MacDonald and Nagayasu, 2000, and Chortareas and Driver, 2001). Strictly speaking, however, these conditions are not LOOP tests, as they abstract from the potentially relevant role played by exchange rate and default risk. Note that, in the case of DRs, the price difference between the two stocks is not affected by expected exchange rate fluctuations, as arbitrage takes place almost immediately. This contrasts with interest rate parity conditions, which look at a much longer horizon.

¹²Closed-end funds cannot be redeemed for the underlying shares, impeding perfect arbitrage. This introduces a distinction between the fund and the underlying portfolio, which is behind the persistent closed-end fund premium. This feature of closed-end funds contrasts with the case analyzed in this paper, in which full arbitrage can be easily performed and a much smaller price divergence is found. Many papers have been written on the closed-end fund puzzle in the U.S.; see for example Lee et al. (1990 and 1991). Other papers focus on international closed-end funds, such as Frankel and Schmukler (1998 and 2000) and Levy Yeyati and Uribe (2000).

¹³An example is Royal Dutch/Shell, which has two shares traded in different markets (Royal Dutch in Amsterdam and Shell in London). It is one firm, but as cash flows are split unevenly, the market value of Royal Dutch must theoretically be 1.5 higher than that of Shell. However, in practice, even though arbitrage is possible, the market value of both stocks has fluctuated far above and below the theoretical difference. One partial explanation for this phenomenon is that Royal Dutch was for a long time a member of the S&P 500 index, while Shell was not, implying that index funds tracking the S&P500 were forced to buy Royal Dutch, even when it was more expensive (Lamont and Thaler 2003).

¹⁴The same should apply to temporary non-zero premia due to differences in trading hours between the domestic and the U.S. stock market.

¹⁵Appendix Table 1 reports the companies that are included in the respective portfolios and the period for which the premium is calculated. Note that only a very limited number of stocks traded in the early 1990s. In the vast majority of countries, firms did not cross-list through ADRs prior to 1994 or even later.

¹⁶On the other hand, when there is trading in both markets during the day, the cross-market premium for that day should closely reflect the contemporaneous transaction costs.

¹⁷Asynchronous trading hours always present a problem when studying comovements of equity prices in different countries and are dealt with in different ways. For example, Bracker et al. 1999 use leads and lags to account for asynchronous trading when studying the comovement of daily returns of ADRs and their underlying stocks. Karolyi and Gagnon (2004) use as control variable the number of time-zones that separate markets when testing whether the return differential between the underlying stocks and the ADRs differs from zero. Pasquariello (2007) uses weekly returns instead of daily returns to limit the impact of asynchronous trading. Other studies (e.g. Yang, 2007) use open and close prices to account for asynchronous trading hours.

¹⁸Note that the premium would gradually decline in absolute value but would not necessarily revert to zero, as arbitrage ceases as soon as the premium is within the band.

¹⁹The implication of the presence of transaction costs as a cause for the existence of two regimes in the data has been mostly developed by the purchasing power parity literature. For example Sercu et al. (1995) and Michael et al. (1997) analyze real exchange rates and develop a theory suggesting that the larger the deviation from PPP, the stronger the tendency for real exchange rates to move back to equilibrium.

²⁰Note that in our model we implicitly assume that the residuals are the same in both regimes. As a result, we can estimate the LLR of the TAR in the same way as the LLR of the AR model and do not need to divide the likelihood function in two parts, one using the residuals of the inner and another one using the residuals of the outer regime, as done by Obstfeld and Taylor (1997). In fact, using this partitioned likelihood function increases the likelihood of rejecting the AR model in favor of the TAR model when residuals are not normally distributed.

²¹In the paper, we estimate a different TAR for no-control and control periods, as convergence of a regression for all periods with some shift parameter to account for the regime change (a priori, a natural alternative) would be extremely difficult and imprecise.

²²For all stocks included in our sample, the mean of the absolute value of the premium on non-trading days exceeds the one on trading days.

²³Note that in the case of Korea estimates are only available for two stocks. As explained in the next section, this is caused by the fact that the remaining four stocks in the portfolio were subject to capital controls over the entire sample period.

²⁴A regression-based analysis could potentially be applied to determine which country-specific and firm-specific characteristics impact the extent of financial integration. Nevertheless, the limited number of countries prevents us from doing meaningful estimations. In the next section, we exploit, however, our large firm-level variation to show how measures of integration are related to stock liquidity.

²⁵For brevity, not all estimated parameters were included in Appendix Table 2. However, we find that, as expected, in almost all cases β_in is not significantly different from zero, providing an indication that inside the band of no-arbitrage the premium follows a random walk. Furthermore, the estimated sum of the ARCH and GARCH parameters lies between 0.90 and 0.99, with a value of 0.95 for the majority of stocks.

²⁶Note that transaction costs are likely to be non-linear (e.g., large transactions command proportionally smaller fees). However, there is a priori no reason to expect that the average trade size of illiquid stocks should be smaller than that of more liquid stocks – if they were, this would add to the liquidity premium.

²⁷See Levine and Schmukler (2006) for alternative measures of liquidity and their close relation with value traded.

²⁸These results are robust to including market capitalization of the stock as an additional control variable.

²⁹As was previously the case, the TAR model was estimated using only contemporaneous trading days.

³⁰Again, results are robust to including market capitalization as an additional control variable.

³¹We only look at controls that directly affect the possibility of arbitrage when they actually restrict the movement of capital across borders. It is difficult to control for the expectations of future capital controls, but given that arbitrage is very rapid, we believe this aspect should be negligible for our computations. For example, the cross-market premium in Argentina became positive only when the country restrictions on capital outflows were actually imposed, even though they were largely anticipated (Schmukler and Serven, 2002).

³²In practice, de jure capital controls create price differences only when they are de facto binding (the crossmarket premium identifies those cases). Otherwise, their presence is de facto immaterial.

³³While capital controls are imposed and lifted with varying financial conditions, they do not seem to be endogenous to the behavior of the cross-market premium. Governments have tended to impose capital controls on outflows to reduce capital flight, and controls on capital inflows to prevent exchange rate appreciations, rather than as a response to stock market fluctuations. Moreover, as Appendix Table 3 shows, controls are not always imposed (or lifted) around crises.

³⁴In fact, the URR was set to zero, but the mechanism was left in place until it was finally eliminated in 2002.

³⁵See the Financial Supervisory Service’s Regulation on Supervision of Securities Business, Article 7-9.

³⁶For Korea, the statistics are derived from the average premium of the unrestricted stocks (no-control period) and the average premium of the restricted stocks (control periods).

³⁷The differential impact across countries of controls on the magnitude of the premium can be the result of several factors, among them the exact type of control. However, a thorough assessment of the precise drivers is beyond the scope of this paper.

³⁸Estimating both thresholds simultaneously in a precise way is exceedingly difficult. Given the variations in the data during the control periods and the length of the time series, several of the models would fail to converge. More critically, on theoretical grounds, we expect only one band to vary when controls on capital inflows or outflows take effect; there is no reason for the other band to be different. Note, however, that the band is not imposed to be asymmetric; the estimated band could be equal to the band estimated during the no-control period.

³⁹For Korea, we cannot make a comparison between no-control and control periods on a stock-by-stock basis as the restricted stocks have been restricted over the whole sample period, while the group of unrestricted stocks did not experience a period of controls. Furthermore, the TAR model cannot be estimated for the stocks in the portfolio of Venezuela due to the limited number of contemporaneous trading days in the control period.

⁴⁰The results for individual stocks in each country are comparable. The estimations are available from the authors upon request.

⁴¹To keep the number of stocks constant over the sample period, we only use stocks that were traded over the entire sample period (2000-2007). These are BBVA Banco Frances, IRSA Inversiones y Representaciones, Petrobas Energia, Telecom Argentina, and Transportadora de Gas del Sur.

⁴²Note that the average AR half-life is different than the one presented in Table 5 [ PDF 45.1KB | 1 page ]. This is because the sample period is different, the number of stocks in the sample is smaller, and the estimated model includes a second control dummy.

⁴³The change in sample period and the smaller number of stocks in the sample explain why the TAR estimates for the no-control and control (on outflows) periods are different from the ones shown in Table 5 [ PDF 45.1KB | 1 page ].

⁴⁴Effectiveness here is understood as the success in producing the desired market segmentation. Whether or not this segmentation is beneficial to the economy is an altogether different question that exceeds the scope of this paper.

⁴⁵The weights are equal to the reciprocal of the standard deviation of the respective variables. Ideally, one would also like to include the change in reserves; unfortunately, these data are not available on a daily frequency for the countries in our sample.

⁴⁶The following rates were used: 7-day interbank rate (Argentina), the bank deposit certificate rate (Brazil), the 30-day CD rate (Chile, Venezuela), the interbank call money rate (Indonesia, Korea, Russia), the 90-day bank deposit rate (Mexico), and the 3-month discount rate (South Africa).

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

Comment(s)

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