165 / 322
臺大管理論叢
第
26
卷第
3
期
165
Table 2 Descriptive Statistics and Unit Root Tests
Variables
Open interest (,000)
Market depth (,000)
Volume
(,000)
OI
EOI
UOI
|
ΔOI
|
DEP
(Bid+Ask)
Bid
Ask
No. Obs.
2,226 2,205 2,205 2,225
2,226 2,226 2,226 2,226
Mean
47,631 47,807 138.82 2,097
323
161
162 63,668
Median
46,186 46,267 500 1,215
308
155
153 50,913
Std. dev.
15,429 14,967 3,701 3,217 140.09 70.84 72.27 39,349
Max.
90,765 89,539 12,475 31,192 1,046
519
700 289,303
Min.
11,636 12,098 -26,148 0
55
23
32
9,539
Skewness
0.20 0.21 -2.83 4.45
0.78
0.82 1.05 0.94
Kurtosis
2.62 2.58 17.14 27.67
3.93
4.19 5.51 3.70
Unit root tests
ADF
-9.67
a
-12.7
a
-46.89
a
-6.70
a
-5.78
a
-3.07 -5.31
a
-6.38
a
Lag number
0
18
0
24
4
14
6
9
Note:
OI
is the open interest,
EOI
is the expected component of open interest,
UOI
is the unexpected
open interest, |
ΔOI
| is the absolute of open interest changes, and market depth is the sum of
order size. All of open interest and volume variables are calculated using two near-month
contracts. The null hypothesis of unit root is tested by the ADF test with a constant term, linear
trend and lag terms suggested by the SBC. The table reports the t-statistics of ADF test, with
superscript a, b and c indicating significance at the 1%, 5% and 10% confidence levels,
respectively.
Summary statistics of market depths are reported in columns 6 through 8 of Table 2.
The ask depth series resembles the bid depth series in the first two moments of the
distribution, but with larger skewness and kurtosis. Since we do not distinguish the buy side
and sell side depth, the bid + ask depth is used in subsequent analysis. Finally, the unit root
tests in the last two rows in Table 2 suggest that each variable is a stationary series (i.e.,
I
(0)
series) except for the bid depth series. The ADF test in every variable rejects the null
hypothesis of one unit root at a 10% significance level.
4.2 Test Results of Open Interest Reflecting Participation
Table 3 presents the test results for implications (1) and (2) of Hypothesis 1 using
Model (1) and (2), respectively. For the results of Model (1), the level of open interest is
positively related to trading volume (
VOL
), depth (
DEP
), and Amivest ratio (
Amivest
) and
negatively associated with the Amihud illiquidity ratio (
ILLQ
). All coefficients of
OI
are
statistically significant except the regression for Amivest ratio. Results suggest that the
increase in open interest is accompanied by higher trading volume, greater depth provision,