relationship between return and volume change. Some of important observations can be summarized as follows:
First, for a relatively short time window, the lagged return is more powerful than the lagged volume change in predicting the return in short-term, and return tends to be reversed. The lagged volume can be a good predictor for the return when it experiences a substantial decrease. Then, the return tends to be negative.
Second, when the information collecting period extends to two weeks, the pattern of the forecast return is much more consistent with the return of previous or next week than that of one week. It implies that using filters based on the lagged return and the volume change in two previous weeks is more powerful to predict the future return than those in previous week.
Third, in the case where the filters based on the lagged return and the volume change are calculated in previous two days, when there is low trading volume, the return tends to be reversed; and when there is high trading volume, the return tends to experience reduced reversal. When the calculations are based on information of previous two weeks, simultaneous increases in the lagged return and the volume change tend to be followed by increase in the
‧
return. These two patterns can be found from the samples spanning 1993-2002, but cannot from the samples spanning 2003-2012. We conjecture that Taiwan stock market is likely to be more dominated by liquidity trading than information trading in recent decade.
The pattern of the return for the combination of the lagged return and the volume change is unclear and, sometimes chaotic. This suggests that the evidence found does not correspond to implications of any aforementioned theory perfectly.
Conrad, Hameed, and Niden (1994) show that if small capitalization stocks experience high growth in volume, their returns will display reversal. If the stocks experience decreases in volume, a higher chance for continuation in their returns can be observed. Their result yield support for Campbell, Grossman and Wang (1993)’s model. Thus, there might be a difference in the relationship in the return and the volume between large and small capitalization stocks. A natural way to extend the research is to investigate the cases with small capitalization stocks. But the challenge of the extension will have to deal with the bid-ask bounce problem.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table I Results for the case where profits are calculated one week after the strategy-formed-day, and lagged return and volume change are calculated in the previous week, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table II Results for the case where profits are calculated two weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous week, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table III Results for the case where profits are calculated three weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous week, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table IV Results for the case where profits are calculated four weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous week, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table V Results for the case where profits are calculated one week after the strategy-formed-day, and lagged return and volume change are calculated in the previous two weeks, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table VI Results for the case where profits are calculated two weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous two weeks, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table VII Results for the case where profits are calculated three weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous two weeks, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table VIII Results for the case where profits are calculated four weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous two weeks, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table IX Results for the case where profits are calculated two days after the strategy-formed-day, and lagged return and volume change are calculated in the previous two days, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table X Results for the case where profits are calculated four days after the strategy-formed-day, and lagged return and volume change are calculated in the previous two days, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table XI Results for the case where profits are calculated six days after the strategy-formed-day, and lagged return and volume change are calculated in the previous two days, 1993-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table XII Results for the case where profits are calculated two weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous two weeks, 1993-2002
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table XIII Results for the case where profits are calculated two weeks after the strategy-formed-day, and lagged return and volume change are calculated in the previous two weeks, 2003-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table XIV Results for the case where profits are calculated two days after the strategy-formed-day, and lagged return and volume change are calculated in the previous two days, 1993-2002
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table XV Results for the case where profits are calculated two days after the strategy-formed-day, and lagged return and volume change are calculated in the previous two days, 2003-2012
*,**,*** The null hypothesis is rejected at the 10%, 5%, 1% levels respectively.