Page 149 - AJWEP-v22i2

Asia’s water scarcity challenge

3. Data and methodology The method is adapted from Khan et al., with data
43
sourced from the World Development Indicators. 41
The present study adopts a methodological approach
designed to ensure the replicability and clarity of the The study focuses on 39 Asian countries as the unit of
research techniques. To address endogeneity issues analysis, chosen from a total of 48 countries in the region
frequently encountered in environmental and resource based on data availability. The panel dataset spans
management research, the analysis employs dynamic 26 years, covering the period from 1996 to 2022. The
panel data with the GMM estimator. GMM is particularly countries are categorized into subregions as follows:
well-suited to capturing dynamic relationships, as • Eastern Asia: China, Japan, Mongolia, North Korea,
it accounts for unobserved heterogeneity, lagged and South Korea
dependent variables, and instrumental effects. Recent • Southern Asia: Afghanistan, Bangladesh, Bhutan,
studies by Arellano and Bond , Blundell and Bond India, Iran, Maldives, Nepal, Pakistan, and Sri
40
39
confirm the effectiveness of this method for panel Lanka.
data analysis. In line with this approach, the following • South-Eastern Asia: Brunei Darussalam, Cambodia,
section details the variables employed in the model: Indonesia, Laos, Myanmar, Malaysia, Philippines,
Singapore, Thailand, Timor-Leste, and Vietnam
(i). Dependent variable • Western Asia: Armenia, Bahrain, Cyprus, Georgia,
• WS (indicating climate risk): Measured as water Jordan, Iraq, Israel, Kuwait, Lebanon, Oman, Qatar,
productivity – total gross domestic product (GDP) Saudi Arabia, Syria, State of Palestine, Turkey,
(constant 2015 US$) per m of total freshwater United Arab Emirates, and Yemen
3
withdrawal – sourced from the World Development • Central Asia: Azerbaijan, Kyrgyzstan, Kazakhstan,
Indicators. 41 Tajikistan, Turkmenistan, Uzbekistan.

(ii). Independent variables The study utilizes a regression model to explore the
• RQ: Captures the government’s ability to establish relationship between governance indicators and water
and enforce robust policies and regulations that governance across Asia. This model, referred to as the
support private sector development. Values range Good Governance and Water Resources Model or the
from −2.5 to 2.5, expressed in standard normally Water Governance Model, is grounded in theoretical
distributed units, obtained from the World relationships established in the present research. It
Governance Indicators 42 aims to examine how governance indicators influence
• Government effectiveness (GEF): Reflects the water governance practices in the region. The model is
quality of public services, civil service independence specified as follows:
from political influences, and the quality of policy
formulation and implementation. Also expressed in WS = ∝ + ∝ WS + ∝ GEF + ∝ RQ + ∝ REC + ∝
4
1
t-1
t
t
0
5
t
t
3
2
standard normally distributed units (−2.5 to 2.5), POPG + ∝ CROP + ∝ AGLD + ∝ CLF + μ (I)
t
8
t
7
t
6
t
t
sourced from the World Governance Indicators. 42
Where WS, GEF, RQ, REC, POPG, AGLD, and
(iii). Controlled variables CLF retain their previously defined meanings, and crop
• REC: The share of renewable energy in total final production (CROP) denotes CROP.
energy consumption, sourced from the World Dynamic GMM is frequently applied by
Development Indicators 41 economists for parameter estimation in changing
• POPG: Measured as the annual percentage change panel data simulations. It proves particularly
39
in population, sourced from the World Development valuable in addressing challenges such as unobserved
Indicators 41 heterogeneity, endogeneity, and serial correlation
• AGLD: Represented by the percentage of land area within panel datasets. A commonly used variant is the
designated for agriculture, sourced from the World two-step differenced GMM estimator, which involves
Development Indicators 41 differencing the data to eliminate individual-specific
• Climate financing (CLF): Constructed using effects and generate valid instruments for accurate
principal component analysis, this index estimation. This estimator operates in two phases.
incorporates weighted components of foreign direct In the first phase, data differencing removes time-
investment inflows, REC, carbon damages, and invariant unobserved heterogeneity, thereby mitigating
trademark application (direct resident) expenditure. endogeneity concerns. In the second phase, the system
Volume 22 Issue 2 (2025) 143 doi: 10.36922/AJWEP025090057