Table 4.
Bounds test for cointegration.
Month
Dependent Variable
F-Statistic
Asymptotic Critical Values (n = 1000)
Finite Sample Critical Values (n = 60)
I(0)
I(1)
I(0)
I(1)
Antiquities Looting
6.249452
3.1
3.87
3.288
4.07
Armed Conflict
4.326421
3.1
3.87
3.288
4.07
Quarter
Dependent Variable
F-Statistic
Asymptotic Critical Values (n = 1000)
Finite Sample Critical Values (n = 70)
I(0)
I(1)
I(0)
I(1)
Antiquities Looting
15.2277
3.1
3.87
3.243
4.043
Armed Conflict
9.499185
3.1
3.87
3.243
4.043
Note: Null Hypothesis is that there is no long-term relationship. F-Statistic must exceed the critical value for I(1) to
reject the null. All critical values (asymptotic and finite sample) are reported for a 95% significance level.
The error correction form of Autoregressive Distributed Lag Models used for these analyses
distinguishes between long-term permanent relationships (either positive or negative) and short-term
relationships that fluctuate depending on the context. Regardless of the short-term fluctuations, it is
the long-term relationship that is of interest for this hypothesis as it assess whether a permanent
relationship exists and the direction of the relationship. Table
5
shows the long-term relationships
for both the month and quarter analyses. The results suggest that in the long-term, an increase in
antiquities looting will lead to a 52.16 increase in the number of armed conflict incidents and that
7
Statistically, the presence of cointegration indicates that two time series are jointly stationary, which is necessary in order to
interpret any findings (
Engle and Granger 1987
).
Arts 2018, 7, 22
15 of 26
an increase in armed conflict will lead to a negligible (0.0049) increase in the number of antiquities
looting incidents.
If a long-term, permanent relationship exists between antiquities looting and armed conflict,
it should exist in both the monthly and quarterly analyses; however, as seen in Table
5
, a positive
long-term relationship appears to only be present in the quarterly analysis. This finding should not
be interpreted as evidence that a long-term relationship does not exist. Instead, as discussed above,
it is more likely that the monthly time period (2010–2014) is not long enough to detect a long-term
relationship (
Marcellino 1999
;
Pierse and Snell 1995
). Overall, the results of the cointegration and
long-term analyses suggest a long-term relationship between antiquities looting and armed conflict
and that it is likely a positive relationship if assessed over a long period of time (e.g., 1997–2014).
Dostları ilə paylaş: |