Elsevier

Journal of Empirical Finance

Volume 33, September 2015, Pages 243-262
Journal of Empirical Finance

The interaction between foreigners' trading and stock market returns in emerging Europe

https://doi.org/10.1016/j.jempfin.2015.03.011Get rights and content

Highlights

  • The first comprehensive study of foreigners' trading in European emerging stock markets using novel data.

  • Risk appetite and global emerging-market returns are significant global drivers of net foreign flows.

  • Foreigners negative-feedback-trade with respect to local returns at the monthly frequency.

  • Negative-feedback-trading asymmetry: foreigners sell after increases, but do not buy after declines.

  • Net flow persistence decreases with local return volatility.

Abstract

This paper presents the first comprehensive study of foreigners' trading in European emerging stock markets, using complete data compiled at the destination. We also compare European results to Asia, to which available evidence is mainly confined. The findings provide new insight on foreigners' trading: in addition to rebalancing, risk appetite and global information are global drivers of net foreign flows. Foreigners' negative-feedback-trade with respect to local returns at longer horizons. They do so with an asymmetry: they sell following increases but do not buy following declines. Net foreign flow persistence decreases with local return volatility.

Introduction

As international investors are perceived to have a large impact on host emerging markets and, at the same time, to herd and positive-feedback-trade, the interaction between foreign investors' trading and emerging equity market returns has been a main focus in the international finance and microstructure literatures. Empirical research on the joint dynamics of foreign investors' trading and emerging equity market returns has examined: i) the global (push) factors that drive equity portfolio flows, ii) the positive feedback trading, which has been considered a symptom of foreign investors' informational disadvantages, and iii) the price impact and forecast ability of foreign flows. However, extant empirical evidence is mainly confined to Asian markets (Choe et al., 1999, Griffin et al., 2004, Richards, 2005, Dvořák, 2005, Choe et al., 2005, Bailey et al., 2007, Samarakoon, 2009; among others) due to data availability. Emerging Europe, where the role of foreign investors is much more pronounced, has been surprisingly neglected.

This paper provides the first comprehensive study of foreigners' trading in European emerging markets, using complete foreign flows data compiled at the destination. By comparing the results for European and Asian emerging markets, we provide a critical out-of-sample assessment of the extent to which available empirical evidence dominated by Asian markets can be generalized. We also contrast the analysis at the monthly frequency and daily frequency, where daily data are available. These, in turn, allow a comprehensive out-of-sample assessment of theories of international investor behavior (Albuquerque et al., 2007, Albuquerque et al., 2009, Barron and Ni, 2008, Brennan and Cao, 1997, Dvořák, 2003, Griffin et al., 2004; and Hau and Rey, 2004, Hau and Rey, 2006). The analysis yields a number of interesting findings that are new to the literature.

European emerging equity markets differ from their extensively-studied Asian counterparts in several important respects: first, foreign investor participation is much higher. For example, in recent years the ratio of market capitalization (under free float) held by foreign investors has ranged between 60 and 80% in Turkey, Hungary, Czech Republic, around or above 50% in all countries in our sample with few exceptions, whereas the same ratio is around 25–35% in Asian markets (was even much lower in periods covered by major papers in this literature). Nonresidents account for nearly half of the traded value in Greece, Poland and many others, whereas their share is around 15–25% of the traded value in Asian markets where local individual investors have a larger role. In Richards' (2005) analogy, the “big fish in small ponds” is actually in emerging Europe. Second, foreigners' trading is free of partial restrictions in European emerging markets: all countries in our sample receive the top rating in MSCI's foreign investor accessibility review, whereas Asian markets, to which available empirical evidence is mainly confined, are still lower-rated (MSCI, 2011). Partial restrictions on foreigners' trading and stepwise elimination of them may have clouded foreigners' true trading patterns. Third, European markets completed their liberalization process earlier, therefore our results represent foreigners' natural trading behavior free of post-liberalization effects documented by Bekaert et al. (2002). Given the second and third points, out-of-sample evidence is needed to obtain generalized conclusions on foreign investor behavior. For these reasons, European emerging markets offer the most suitable venue to study the interaction between foreigners' trading and emerging stock market returns. Finally, European emerging economies are those which are most dependent on foreign capital inflows. Large external deficits and host economies' dependence on foreign capital make understanding and predicting international capital flows' behavior more crucial.

While several panel studies using data from one custodian or one source country cover some European emerging markets,1 such data do not include all foreign investors and may be unsafe to generalize. In particular, US Treasury International Capital (TIC) data set has been employed by major studies in this literature. The correlation between our net foreign flows compiled at the destination and those derived from the TIC data set is merely + 0.150 for Turkey, + 0.192 for Greece, + 0.124 for Hungary and + 0.268 for Poland. The differences between the two data sets are due to: i) TIC data represent only US investor flows whereas our destination-compiled data cover all foreign investors, and ii) TIC data contain biases as discussed by Griever et al. (2001) and Bertaut et al. (2006). The discrepancy between TIC data and destination-compiled complete data is particularly pronounced for European markets due to the role of London as a financial hub. Therefore, there is a gap regarding evidence on foreign investor behavior in emerging Europe, an important geography that can offer unique insights on foreign investors' trading.

The main reason behind this surprising gap has been the lack of foreign flows data.2 This study overcomes the data availability obstacle by combining different sources to create a complete and exact (where possible), destination-compiled emerging-Europe data set of monthly foreign flows with a comprehensive coverage of countries. We also employ the sole available daily data from a sizeable European emerging market (Turkey). The comprehensive sample of structurally similar economies, which are highly integrated to global financial markets, enables a well-suited reassessment of theories of foreign investor behavior. We provide out-of-sample evidence on the following issues:

First, we characterize the global (push) factors. Theories of rebalancing (Bohn and Tesar, 1996, Griffin et al., 2004, Kodres and Pritsker, 2002, Kyle and Xiong, 2001) predict a positive relationship between developed (home) market returns and flows towards emerging (host) markets as a result of wealth and portfolio-balance effects. Extant empirical evidence is consistent with a positive relationship (Griffin et al., 2004, Richards, 2005, Ülkü and İkizlerli, 2012).3 The current paper provides a break-down of the positive relationship between global market returns and flows towards emerging markets into portfolio rebalancing, time-variation in risk appetite, and global information. As emerging equities constitute a high-risk asset class, flows towards them may respond to changes in risk appetite. Similarly, as foreign investors may be timing their investments based on information about global macroeconomic conditions that will affect emerging markets as a whole, their trading may be correlated with global emerging stock index returns which serve as a proxy for expectations on future global macroeconomic activity. The idea of global private information driving foreign flows is elaborated on by Albuquerque et al. (2009); the current study provides further empirical evidence on the hypotheses developed there. Cross-emerging market rebalancing (Kyle and Xiong, 2001) can also drive foreign flows. By employing indicators of time-variation in risk appetite (the credit spread, or the volatility index, VIX) and returns of a global emerging market index, in addition to a global developed market index, within appropriate structural identification schemes, we decompose net foreign flows' response to world returns. Our results point to a significant role of global emerging-market-related information and somewhat of risk appetite, in addition to developed (home) market returns, as drivers of net foreign flows.

Second, early theory on international equity portfolio investments characterizes foreign investors as “uninformed positive feedback traders” with respect to local returns [Brennan and Cao (1997), Griffin et al. (2004); Albuquerque et al. (2007) differ by emphasizing the heterogeneity within the group of foreigners]. The intuition stems from the premise that access to local information is costlier for foreign investors, thus public information arrivals will result in bigger revisions in foreign investors' valuations (Brennan and Cao, 1997). Thus, local investors' cumulative information advantage4 would lead to a positive relation between current net foreign flows and contemporaneous and lagged local returns. A positive contemporaneous relation on low-frequency data, however, does not necessarily imply intra-period positive feedback trading; it may be consistent with contemporaneous price pressure and/or intra-period forecast ability of net foreign flows, too. In contrast, Dvořák's (2003) model predicts a negative contemporaneous relationship between local returns and net foreign flows based on local investors' marginal information advantage. Hau and Rey, 2004, Hau and Rey, 2006 model predicts a negative relationship for a different reason: differential equity-market performance induces portfolio rebalancing to bring exchange rate exposure back to previous levels (i.e., selling following host market positive return shocks). Extant empirical evidence is mostly consistent with a positive contemporaneous relationship, and a positive response of net foreign flows to past local returns, both at the marketwide and individual stock level and irrespective of the frequency of data.5

The current paper provides a major out-of-sample assessment of previous findings. We confirm a positive contemporaneous relationship in most cases, but find a negative response of net flows to lagged local returns at the monthly frequency. A negative response of net foreign flows to lags of local returns is a new finding. It also applies to Asian markets, hence is a general feature (previous findings of positive feedback trading on Asian markets are not valid at the monthly frequency). Net foreign flows' response to global returns is always positive (at both monthly and daily frequency), without any subsequent reversal. Our result of ‘negative feedback trading in all but Eurozone countries’ may be consistent with Hau and Rey, 2004, Hau and Rey, 2006 risk-rebalancing theory, which hypothesizes currency risk to be the motivation for rebalancing.6 Finally, we uncover an asymmetry in negative feedback trading: foreigners' negative-feedback-trade following only positive local returns, not negative ones. This asymmetry parallels a recent finding by Curcuru et al. (2011) employing TIC data corrected for financial center bias.

Third, theory predicts net foreign flows to be persistent (Albuquerque et al., 2007). Persistence may follow from heterogeneity within the group of foreign investors in terms of the speed of accessing new information and acting on it. Empirical literature documented strong evidence of persistence in net foreign flows, especially towards emerging markets (e.g., Froot and Donohue, 2002). However, early empirical evidence on persistence may have been driven by post-liberalization effects documented by Bekaert et al. (2002). The comprehensive European sample employed in the current study, which is free of post-liberalization effects, enables us to assess whether persistence of net foreign flows is a universal characteristic, as in the model of Albuquerque et al. (2007), or was just a temporary phenomenon driven by post-liberalization effects. Results show that a certain degree of persistence in net foreign flows is enduring. The cross-country variation in persistence points to a new finding: persistence in net foreign flows seems to decrease with the volatility of local returns.

Finally, research has also focused on foreign flows' price impact, and their ability to forecast future local returns of the host market. An assessment of the forecast ability would clarify whether the contemporaneous positive relationship mentioned above is due to foreigners' cumulative information disadvantage or marginal information advantage/superior processing of available information. Previous research documents a limited degree of short-horizon forecast ability at the market-wide level (Dahlquist and Robertsson, 2004, Froot and Ramadorai, 2001, Froot et al., 2001, Griffin et al., 2004, Kamesaka et al., 2003, Richards, 2005),7 which is mainly due to persistence in flows (i.e., current flows predict future flows which are positively associated with contemporaneous returns). Many studies conclude that foreign investors are sophisticated traders (Chen et al., 2009), but lack local information (Dvořák, 2005).8 Our data enable an out-of-sample assessment of previous conclusions on net foreign flows' price impact and forecast ability. Results point to a modest degree of short-horizon forecast ability of net foreign flows in emerging Europe, partly driven by the persistence in net foreign flows, but no symptoms of uninformed behavior and no evidence of pure price pressure.

In addressing these issues, additional insight can be gained by using daily data. However, foreign flows data at the daily frequency are not available in any European emerging market.9 In this paper, we use the sole substitute for daily foreign flows data from emerging Europe: the Central Registry Agency of Turkey (CRAT) reports marketwide ownership by resident and nonresident investors on a daily basis, from which we derive net foreign flows. For comparison purposes, we also employ daily data from the three Asian markets mentioned above. Combining and contrasting monthly and daily results offer new insight: while ‘permanent price impact along with a modest forecast ability of net foreign flows’ holds on both monthly and daily data, foreigners display positive feedback trading at the daily frequency but negative feedback trading at the monthly frequency, in both European and Asian markets. Using a model with 50 daily lags, we show that the initial positive response starts to reverse from around day 5–20 in all markets. Thus, the new finding of this paper, negative feedback trading, is not specific to European emerging markets, but specific to the longer horizon. The initial response is positive regardless of the sign of local return, but the subsequent reversal beyond day 5–20 is mainly confined to responding to positive local returns. This signifies a major refining of previously-established stylized facts.

In sum, the paper's contribution stems from: 1) employing novel data that enables a major out-of-sample study of foreign investor behavior on an important, but previously unexplored geographical region, that offers a more suitable venue to test theories of foreign investor behavior, and 2) documenting new stylized facts and refining previous findings on the interaction between stock market returns and net foreign flows, thus reshaping the set of empirical results in the literature with new insights. The article is organized as follows. Section 2 describes the data used in this study and the methodology employed. Section 3 presents the results in subsections organized by research question. Section 4 summarizes the conclusions.

Section snippets

Data

Our monthly data cover eight European emerging markets: Turkey, Greece, Hungary, Poland, Czech Republic, Slovenia, Romania and Bulgaria.10

Results

We present results by depicting cumulative IRFs.16 In IRF graphs to follow, the solid line in the middle tracks the response to a 1-standard deviation shock. The dashed lines around it represent bootstrapped 90% confidence bands. Statistical significance for cumulative (lagged cumulative) response is

Conclusions

This paper provides out-of-sample empirical evidence on the interaction between international investors' trading and emerging stock market returns, employing novel data from an important but previously unexplored region. The comprehensive empirical analysis, comparisons to Asian results and contrasting the monthly and daily results enable us to assess the extent to which previous conclusions in the foreign equity flows literature can be generalized, and to uncover new stylized facts.

First, risk

Acknowledgments

I thank László Varga (Hungarian National Bank), Tomasz Wisniewski (Warsaw Stock Exchange), Ewa Lawskowska (National Bank of Poland), Czech National Bank, Hellenic Exchanges Group, Central Registry Agency of Turkey, Austrian National Bank, Budapest Stock Exchange, Oksana Kondratova (RTS Stock Exchange), Gordana Miskulin (Zagreb Stock Exchange), MSCI-Barra, Bulgaria Stock Exchange, Taiwan Stock Exchange and US TIC for their help in compiling and cleaning the data, Ebru Demirci for research

References (43)

Cited by (0)

View full text