Próxima Página| Próxima Página |
| Página Anterior |
Índice2.1 Ibovespa overnight returns
None of the Asian markets included in the MSEMF-ASIA index have overlapping trading hours with the Brazilian markets. If the information revealed in East Asia during the night (in the western hemisphere) had any effect on markets in Brazil, it would do so right after the opening bell rings. This led us to define a series of Ibovespa overnight returns (RIBOV-ON) and test it for causality effects with the returns of the Asian index. The Ibovespa overnight returns are defined as:
where Ibovespat(o)
is the Ibovespa level at opening time; and
Ibovespat-1(c) is the closing level of the Ibovespa
on the previous day.
The RIBOV-ON
series comprises the close-to-open returns, thus capturing the price moves
caused by new information generated overnight. Part of this information
might be revealed through the Asian markets daytime returns. A statistically
significant impact of these returns on RIBOV-ON, in
the context of the Granger test, is interpreted as a signal of a spillover
effect from East Asia to the Brazilian market. Figure 6 shows both series
of returns.
Due to nonsynchronous trading of the various stocks in the market index, its opening quote may not accurately reflect the true underlying market price. The Ibovespa is currently highly concentrated on a few stocks. Although this would in principle reduce the nonsynchronous trading problem in the 1997 sample, the risk remains of spurious lagged correlations being detected. To avoid this problem, we pick a proxy for the opening quote at a later time, after the official opening of the exchange, when trading would have reached a "steady state". Specifically, Ibovespat(o) is defined as the Index quoted at 11am (one hour after the official opening) from April 8th to October 27th and at 12:30pm from October 28th on (half an hour after the official opening). On that day, the acute market instability prompted regulators and exchange officials to postpone opening time to 12pm. The new arrangement lasted until January 2nd 1998, when it returned to the old norm (10am to 5pm, with a 1-hour lunch break at 1pm). Available intraday data limited the series to 184 observations.
| Próxima Página |
| Página Anterior |
Índice