175 / 322
臺大管理論叢
第
26
卷第
3
期
175
Note:
The
p
and
q
respectively stand for the AR and MA terms of the ARIMA (
p
, 0,
q
) model. The AIC
and SBC are information criterions proposed by Akaike (1974) and Schwarz (1978) respectively.
The residual diagnostics report the ACF, PACF and the Ljung-Box Q-statistics. We show that the
k lags of ACF and PACF of residual which are not within these bounds and it is significantly
different from zero at the 5% significance level approximately. The column of Q-statistics
provides the lag number of the Q-statistics and their corresponding p-value in the parenthesis.
The last model (Model (8) in bold text) is the one used in the study for the OI decomposition.
This model has small value of AIC and SBC and no serial correlation up to 60 lags. For the three
models (Model (5), (6) and (7) in italic) with even smaller AIC and SBC, the residuals are not
white noise according to the Q-statistics.
5. Conclusions
The study provides interpretation to the information contents of open interest, a variable
serving for many proxies in literature. We analyze the index futures traded in Taiwan to test
three hypotheses regarding the information contents of open interest. Specifically, we test
whether open interest reflects market participation, hedging demand and divergence in
traders’ opinions.
The results show that increase in open interest is accompanied by higher trading
volume, greater depth provision, and lower market impact costs. The results are consistent
with Bessembinder and Seguin (1993) that participation, proxied by open interest, enhances
market liquidity. In addition, both expected and unexpected open interest are positively
related to liquidity proxies, supporting the findings of Martell and Wolf (1987), Moser
(1994) and Bessembinder and Seguin (1993). Our findings of the close linkage between open
interest and liquidity variables support the hypothesis that open interest reflects the degree of
market participation.
Our empirical results also corroborate with the hypothesis that open interest
representing demand for hedging. Regression results show significantly positive
relationships between open interest and three spot volatility proxies. The expected
component of open interest moves directly with index volatility, indicating that greater spot
market volatility motivates hedging demand, which adds open interest on futures markets.
The evidence conforms to the viewpoints of Chen et al. (1995), Chang et al. (2000) and
Aguenaou et al. (2011), who assert that open interest reflects hedging demand.
Our results agree with the hypothesis that open interest represents the divergence of
traders’ opinions. We find that both increments and decrements in open interest have
significant but asymmetric impacts on the change in volume. The effect of open interest
increases on trading volume is greater than the effect of open interest decreases on trading