RDP 2004-05: Big Fish in Small Ponds: The Trading Behaviour and Price Impact of Foreign Investors in Asian Emerging Equity Markets 5. Conclusion

This paper has used new data for the daily net purchases of foreign investors in six Asian equity markets to examine the relationship between returns and investor flows. The use of precise daily data for the purchases of all foreign investors has enabled the discovery of two new ‘stylised facts’ about the role of foreign investors that have not been apparent in earlier work using less comprehensive or lower frequency data.

First, the trading decisions of foreign investors in the equity market studied here are substantially influenced by recent returns in global equity markets in addition to returns in the domestic market. The estimates of the relative importance of foreign and domestic returns in explaining flows would indeed suggest that ‘push’ factors on average are at least as important as ‘pull’ factors in explaining flows to these emerging markets. Since foreigners are essentially all institutional investors, this finding presents a very strong example of a form of high-frequency momentum trading by institutional investors (and contrarian trading by individuals), adding to the evidence for this form of trading in other studies using lower-frequency data (e.g., Grinblatt, Titman and Wermers 1995). This adds to the growing body of evidence that investor heterogeneity is an important element in understanding the dynamics of financial markets. What is most surprising about this evidence of positive-feedback or momentum-type investing is its timing: that the trading of foreigners responds so quickly, to price changes that have occurred the previous day or overnight. There is scope for further research into the possible causes of this positive-feedback trading, but it seems reasonable to conjecture that it is the result of some combination of foreign investors using recent returns to extract information about future returns, and sentiment-driven trading or behavioural effects.

The second major result is that there are substantial price movements associated with the trades of foreign investors. In contrast to the work of Froot et al (2001), which suggested that the contemporaneous price impact of foreigners' trades was very small, with prices moving only in the weeks and months following the trades, the analysis here shows that most of the impact occurs on the same day as the trades. The primary reason for the divergence in findings appears to be the use of daily data for the actual trades of all investors, rather than data for a subgroup of foreign investors where the timing of their trades has to be inferred from settlement conventions in each country.^[20] The estimated price impact of foreign inflows is much larger than has been suggested by previous work on emerging markets. However, it is quite similar to the price impacts of trading that are reported in some earlier studies using US data. This suggests that price pressure is a phenomenon that is widespread in financial markets and largely independent of the extent of market development.^[21]

Together, these results suggest that foreign investors and conditions in mature markets have a much larger effect on emerging markets than has been suggested by previous work. However, the combination of trading driven substantially by conditions in other markets and large price pressures from the trading of foreigners raises the possibility that foreign trading can be destabilising in emerging markets. Indeed, the experience of the mid 1990s and then the Asian crisis of 1997 suggests that foreign flows can contribute to both rising and falling prices in emerging markets. These effects may be an unavoidable phenomenon in emerging markets, so the efforts of policy-makers should be directed at ensuring that their markets and institutions are sufficiently strong to be robust to inflows and outflows and the price changes that accompany them.

Footnotes

Differences in methodology would not appear to be able to explain the differences in results. For example, if I replicate the VARs in Froot et al by using their bivariate system with 40 lags, and calculate impulse responses over 60 days, my data still suggest a very rapid impact of flows on returns. In addition, when I replicate the quarterly covariance decomposition in their Table 8, I find a far higher proportion of the total covariance is attributable to the contemporaneous covariance and a much smaller proportion to the covariance between flows and returns 6–60 days later. Since our data are for different periods, it is possible that the differences in results might be due to structural change in the relationships. One possibility is that the period leading up to and including the crisis period was a rather special period for these markets, and it may not be surprising if the subsequent, more normal market conditions lead to more reasonable and intuitive empirical results. [20]

Indeed, by some measures the global foreign exchange market might be considered the most liquid financial market in the world, yet even there it has been noted that estimates of the elasticity of exchange rates to customer order flow are puzzlingly high (Lyons 2001, p 265). [21]