Geography Reference
In-Depth Information
We further validated the time sequence and predictability of the causal linkages
in Fig.
14.7
by using Granger Causality Analysis
29
(Zhang et al.
2011a
). Via the
Granger Causality Analysis, the causal relationship between variables is confirmed
only if the cause precedes the effect in time and the causal series contains special
information that could better explain and forecast the series being caused (Granger
1988
). The causal linkages in Fig.
14.7
boiled down to these relationships: Climate
change
!
bio-productivity
!
agricultural production
!
FSPC; FSPC
!
social dis-
turbance
!
war; FSPC
!
famine
!
nutritional status; FSPC, social disturbance,
war, and famine
!
migration; nutritional status and migration
!
epidemics; war,
famine, and epidemics
!
population; population
!
agricultural production; and
population
!
FSPC. Our Granger Causality Analysis results show that all null
hypotheses of these linkages were rejected (13 linkages with P < 0.01 and 4 linkages
with P < 0.05), implying that causal relationships between climate change and
various human crises are statistically valid (Table
14.7
).
14.4
Discussions
In our studies, we scientifically demonstrated the connection and also the causal
mechanisms between climate change and various human crises in pre-industrial
human societies. In addition, we showed that the societal impact brought about by
global climate change was geographically diversified. In regions where population
pressure and/or agricultural dependency on climate were higher, the climate-crisis
relationship was also stronger and more apparent.
Since 2005, the year in which our first quantitative study about the climate-
war relationship in historical China (Zhang et al.
2005
) was published, there has
been a surge in quantitative empirical large-N studies that explore whether and
how climate change affects human crises, which allows more scientifically robust
and generalizable arguments about the climate-crisis nexus to be generated. For
those large-N quantitative studies that span several centuries or even longer period
(Zhang et al.
2010
; Tol and Wagner
2010
;Wangetal.
2010
), all of their findings
concur with our conclusion about the importance of climate change in affecting
human societies. However, for those large-N studies which use empirical data for the
29
If one factor (
A
) causes another factor (
B
), then changes of A happen first, followed by changes
of
B
, that is the cause precedes the effect. Under this circumstance, Granger Causality Analysis
could be used to test whether
A
which precedes
B
could better help predict
B
by adding
A
. Then,
we consider
A
is the Granger cause of
B
(i.e., the causal relationship between
A
and
B
). Granger
Causality Analysis has been used widely in business, economics, sociology, psychology, politics,
biology, and medicine. It also is regarded as an effective method to verify causal relationships in the
social sciences. Before Granger Causality Analysis, an Augmented Dickey-Fuller test was adopted
to check the stationarity of data. Any nonstationary data were subjected to first- or second-level
differencing. Then regressions were run (by controlling the number of lags) to identify the causal
relation.
Search WWH ::
Custom Search