Comparative water law, policies, and administration in Asia: Evidence from 17 countries
 Conventional wisdom suggests that improving water governance is the key to solving water insecurity in developing countries but there are also many disagreements on operational and methodological issues. In this paper, we build on the work of Saleth and Dinar and surveyed 100 water experts from 17 countries in Asia to compare 19 indicators of water laws, policies, and administration among and within countries from 2001 to 2010. We present the results of our study in a comparative dashboard and report how water governance indicators vary with a country's level of economic development, which ones do not and how and why some indicators change overtime in some countries. We have two main results. First, our initial findings suggest the possibility of water Kuznet's curve, i.e., certain water governance indicators vary with a country's level of economic development. However, more studies are needed given the caveats and limitations of our study. Second, the results have practical value for policy makers and researchers for benchmarking with other countries and tracking changes within their countries overtime. We conclude with implications for a second-generation research agenda on water governance.
 Improving water governance is widely regarded as the key to solving water insecurity problems in developing countries [Rijsberman and Zwane, 2008; Rogers and Hall, 2003; Briscoe, 2009; Hoekstra and Chapagain, 2006; Kashyap, 2004; Saleth and Dinar, 2005; Gopalakrishnan et al., 2004; Biswas and Tortajada, 2010; Asian Development Bank, 2004; Global Water Partnership, 2002].
 However, despite its importance, there remains little consensus amongst water scholars on a number of issues (see Araral and Wang  for a review). First, there is no consensus on the scope, definition, and measurement of water governance, see for example, contrasting definitions from Global Water Partnership , united nations development program (UNDP) Water Governance Facility, Biswas and Tortajada , and Rogers and Hall , among others. The Global Water Partnership  defined water governance as “the range of political, social, economic and administrative systems that are in place to develop and manage water resources, and the delivery of water services, at different levels of society.”
 This definition, however, is problematic because practically the entire literature on water policy, economics, finance, politics, regulation, law, and management would fall under this definition. At the minimum, this definition suffers from a specification problem, i.e., the mechanisms to develop and manage water resources are often not well specified and thus their operational implications for research and governance reform are unclear. We provide an alternative operational definition of water governance in terms of various dimensions of water law, policies, and administration that have been commonly regarded in the literature as important determinants of performance. These include water rights, pricing, decentralization, accountability, integration, private sector participation, user group participation, and organizational basis of water management, among others.
 Second, water governance has largely been studied in terms of disciplinary orientations, i.e., political sociology [Mollinga, 2008], institutions [Pahl-Wostl et al., 2008], institutional economics [Saussier and Menard, 2000; Shirley, 2002], international relations [Konca, 2005], and welfare economics [Rogers and Hall, 2003], among others. As a result, the literature has not evolved into a multi and interdisciplinary agenda despite the fact that water governance should be inherently multidisciplinary in orientation. We address this issue by taking a multidisciplinary approach to water governance by integrating water law, policy, economics, and administration.
 Finally, scholars remain divided on how to approach the study of water governance. Some scholars such as Saleth and Dinar  employ a comparative approach, others use single case studies, while Biswas and Tortajada  propose an alternative approach based on independent and objective case studies of good practices particularly of “the enabling environment and critical factors of success.”
 In this paper, we build on the work of Saleth and Dinar by providing an in-depth and nuanced comparison of 19 indicators of water governance for 17 countries in Asia based on income levels as well as intertemporal analysis within countries from 2001/2002 to 2009/2010. Second, we collected additional 49 survey responses in 2009–2010 to increase the sample size to 100 respondents and strengthen the robustness of Saleth and Dinar's work. Third, we added three countries—Singapore, Uzbekistan, and Mongolia—which were previously not covered in the Saleth and Dinar survey. Fourth, we provide insights to explain the significant changes in water governance practices in selected countries between 2001 and 2010. Finally, we outline the implications of the paper for a second-generation research agenda on water governance.
 The paper is organized as follows. In the next section, we describe the framework, data, methods, and analysis for the study. This is followed by discussion of the findings and analysis and the paper's conclusion and implications.
2. Framework, Data, Methods, and Analysis
2.1. Conceptual Framework and Variables
 We build on the conceptual framework originally developed by Saleth and Dinar , which consists of three dimensions, namely water law, water policy, and water administration. Table 1 summarizes the components and the definitions of these three dimensions of water governance. Most of the variables in our study are ordinal variables, while three are nominal (discrete) variables. The components were chosen to represent the concept of water governance as they have been frequently cited and debated in the literature and in policy discussions [Dinar and Saleth, 2005] as well as being part of the widely accepted Dublin Principles on Water Management. The variables are also amenable to direct policy manipulation, which makes them even more appealing.
Table 1. Conceptual and Operational Indicators of Water Governance Sourcea
|Water law||L1||Legal distinction of different water sources (ORD)||This represents the degree to which varying water sources treated alike or differently by water laws (i.e., surface water and ground water). It is on a scale of 0–10, 10 being “Very Different,” 0 being “Alike.”|
|L2||Format of surface water property rights (NOM)||This indicates the basis of general rights in surface water. The scores center around the following criteria: none, not clear, common or state property, multiple rights, riparian system, appropriative system, correlative system (equal or proportional sharing), and license/permits.|
|L3||Legal accountability of water sector officials (ORD)||This represents the effectiveness of accountability provisions by water laws for water officials. It is on a scale of 0–10, 10 being “Highly Accountable,” 0 being “No Accountability.”|
|L4||Decentralization tendency within water law (ORD)||This illustrates whether or not present laws contribute to centralization and the strength of the tendency of present laws toward centralization. It is on a scale of 0–10, 10 being “Highly Centralized,” 0 being “Highly Decentralized.”|
|L5||Legal scope for private and user participation (ORD)||This represents how favorable the legal provisions for private sector, nongovernmental organization (NGO), and community participation in water development/management are. It is on a scale of 0–10, 10 being “Very Favorable,” 0 being “Unfavorable.”|
|L6||Legal framework for integrated treatment of water sources (ORD)||This indicates the integration level of water laws with other laws on land, forest, and environment. It is on a scale of 0–10, 10 being “Highly Integrated,” 0 being “Fragmented.”|
|Water policy||P1||Project selection criteria (NOM)||This indicates the criteria used in water project selection and how extensively they are applied in irrigation, urban, and multipurpose projects. The scores center around the following criteria: no response, political dictates, equity factors, ecological factors (ECO), benefit-cost ratio (BC), internal rate of return (IRR), and multiple criteria.|
|P2||Linkages with other policies (ORD)||This represents the extent of the influence of other policies on water policy. It is on a scale of 0–10, 10 being “Highly Influential,” 0 being “No Influence.” The linked policies include agricultural policies, energy and power policies, soil conservation policies, pollution control and environmental policies, fiscal policies (structural adjustment), credit and investment policies, and foreign investment and aid policies.|
|P3||Pricing policy (ORD)||This represents the extent of cost recovery by tariffs. It is on a scale of 0–10, 10 being “Full Cost Recovery,” 0 being “Full Subsidy.” The average of domestic, industrial, and irrigation pricing policies is derived.|
|P4||Private sector participation (ORD)||This corresponds how favorable water policy is on private sector participation. It is on a scale of 0–10, 10 being “Very Favorable,” 0 being “Unfavorable.” The scores are averaged across the domains of irrigation, urban domestic use, rural domestic use, and industrial and commercial use.|
|P5||User participation (ORD)||This explains how favorable water policy is on user participation and decentralization. It is on a scale of 0–10, 10 being “Very,” 0 being “Unfavorable.” The scores are averaged across the domains of irrigation, urban domestic use, rural domestic use, and industrial and commercial use in the stages of planning & development and operation & maintenance.|
|P6||Linkage between water law and water policy (ORD)||This represents the extent of the linkages between water law and water policy. It is on a scale of 0–10, 10 being “Tightly Linked,” 0 being “No Linkage.”|
|P7||Attention to poverty and water (ORD)||This represents how well the concerns of the poor are reflected by water policy. It aggregates two components—the existence of such policies and their effectiveness and extent. It is on a scale of 0–10, 10 being “Highly reflected,” 0 being “Hardly reflected.”|
|P8||Finance for water investment (ORD)||This represents the adequacy of funding available for current and future water investments. It is on a scale of 0–10, 10 being “Highly Adequate,” 0 being “Inadequate.” The scores are averaged across the funding for “new Infrastructure,” “utilities repair and O&M,” “irrigation,” and “water resources management.”|
|Water administration||A1||Organizational basis (NOM)||This shows the basis on which water administration is organized. The scores center around the following criteria: on administrative division (geographical basis), on hydrogeological regions, on river basins, and mixture of all.|
|A2||Functional balance (ORD)||This indicates whether or not functional specialization within water administration is balanced. It is on a scale of 0–10, 10 being “Highly Balanced,” 0 being highly “Unbalanced.” The tested functions are planning and design, implementation, financial management, operation and maintenance, rehabilitation and resettlement, environmental monitoring, research, training, and extension, interagency, or departmental relationships|
|A3||Existence of independent water pricing body or apex body (ORD)||This represents the existence of independent bodies for determining water price. It is on a scale of 0–10, 10 being “Existent,” 0 being “Nonexistent.”|
|A4||Accountability and regulatory mechanisms (ORD)||This represents the effectiveness of the accountability arrangements evaluated. It is on a scale of 0–10, 10 being “Highly Effective,” 0 being highly “Ineffective.” The accountability mechanism was analyzed with respect to both within and outside of formal administration.|
|A5||Validity of water data for planning (ORD)||This represents the adequacy and reliability of water data for planning purposes. It is on a scale of 0–10, 10 being “Highly Valid,” 0 being “Invalid.”|
|A6||Science and technology application (ORD)||This indicates the extent to which the following science and technology components are used within water administration: computers, remote sensing and satellite, research and experimental information, modern accounting and auditing techniques, management information systems, geographic information systems, wireless communication, water-measuring technology, computerized dynamic regulation of canals, and water delivery networks. The aggregate score is on a scale of 0–10, 10 being “Very Extensive.” 0 being “Very Low.” The scores are averaged across the technologies specified above.|
2.2. Sample Data and Questionnaire
 Our research data are based on two time periods: the 2001–2002 survey by Saleth and Dinar  and the 2009–2010 survey by this study. Taken together, the respondents of these two surveys encompass 100 water professionals from 17 countries in Asia. The use of expert opinion has been the conventional method for constructing composite indices over the years because objective information is rare or unattainable for qualitative concepts like water institutions. Studies that systematically compare various dimensions of water governance across countries are rare and, therefore, a comparative survey would be valuable.
 While there are many studies on water governance institutions, they have serious limitations: (1) they do not allow for a more systematic comparison across countries of varying levels of economic development; (2) they do not allow for systematic comparison overtime; and (3) they are not cost effective, i.e., we have to pull together close to 1600 data points given the number of indicators and subindicators that we wanted to compare (at least 40) and the number of countries we were comparing (19) across two time periods. Therefore, because of these limitations, the use of comparative survey data from water experts helps address these limitations.
 Popular examples of composite indices using expert judgment include the widely recognized Corruption Perception Index (CPI) by the Transparency International and the governance indicators developed by Kaufmann et al. . Numerous studies show that such qualitative indices exhibit behavioral consistency with their linked “objective” performance measures when they are correlated against each other. This consistency indicates and reinforces the pertinence of such an approach [Kaufmann et al., 2003]. The details of the country coverage and the response frequency per country for the two time periods are shown in the Table 2. The countries that appear in both surveys are Bangladesh, Cambodia, People's Republic of China (PRC), India, Indonesia, Japan, Lao PDR, Nepal, Pakistan, Philippines, Sri Lanka, Thailand, and Vietnam.
Table 2. Country Coverage and Survey Response Counta
|Southeast and South Asia||Bangladesh (4, 2), Cambodia (1, 5), India (9, 3), Indonesia (4, 4), Lao PDR (1, 3), Nepal (2, 4), Pakistan (3, 3), Philippines (3, 5), Singapore (0, 3), Sri Lanka (3, 2), Thailand (1, 4), Vietnam (2, 4)|
|Northeast Asia||Japan (4, 1), Korea (1, 0), China-People's Republic of (5, 2), Mongolia (0, 1), Taipei-Taiwan (1, 0)|
|Central Asia||Uzbekistan (0, 3)|
|Oceania||Australia (6, 0), New Zealand (1, 0)|
 Out of the total 100 survey responses, 51 came from the Asian portion of the 2001–2002 survey. The remaining 49 were collected by this study in 2009–2010. We essentially used the same questionnaire as Saleth and Dinar to enable intertemporal comparison of changes in water governance. The questionnaire used in 2009–2010 can be found in an online survey link (http://www.surveymonkey.com/s/7WVPGRV). The questionnaire used by the Saleth and Dinar can be found in Appendix A of Saleth and Dinar . In the 2001–2002 survey, 48% of the respondents are engineers, 32% are economists, and the rest are either lawyers or social scientists of various kinds (we were unable to isolate the profile for the Asian portion of the Saleth and Dinar sample; as such, the above profile is based on the entire sample).
 In the recent survey, 53% are engineers, 27% are government officers who are not engineers, 8% are economists, 6% are academicians, and the remaining portions are lawyers, businessmen, and information technologists. This trend in profile is more or less consistent with the existing disciplinary composition found in the water sector of most countries [Dinar and Saleth, 2005]. The names of the participants of the recent survey are available with the authors.
 There are three mechanisms to ensure reliability of the survey responses. First, the respondents were selected based on their responsibilities and years of experience in the water sector in their countries. Thus, we have respondents who are managers and leaders of water utilities, regulatory bodies, water apex bodies, water ministries, academics, and the private sector, among others. Second, the responses we obtained came from a pool of expert respondents so standard errors are distributed. Third, we also reported our sample frame and standard deviations so readers can judge the reliability of the responses.
2.3. Methods of Analysis
 We employ four methods of analysis. First, for the 16 ordinal variables in our data set (see Table 1 for coding), we used the mode as our measure of central tendency by country and for each of the survey periods (2001/2002 and 2009–2010). We did not use the mean because it is not a meaningful measure for ordinal variables and the median is not appropriate given the small sample size. Our survey questionnaire originally used a more variable ordinal scale (i.e., 0–100) because the plan was to construct a water governance index. Since we have dropped this plan in favor of simply reporting the raw scores, we decided to normalize the ordinal scale from 0–100 to 0–10 by simply dividing the raw scores by 10 and rounding up to the nearest unit. The results are the essentially the same but our normalization has made it easier to compare the results across countries and within countries overtime. For nominal variables such as water rights (L2), project selection criteria (P1), and organization basis of water administration (A1), we asked respondents to choose among the options given and used the mode for data analysis.
 Second, we report in a comparative table (dashboard) the scores for each of the 19 water governance indicators for both time periods for all of the countries covered in the survey. Third, using these raw scores from step 2 above, we compared the weighted means and standard deviations in 19 areas of water governance among the 17 countries based on income levels (low, middle, and high incomes). The mean of subgroup central tendencies can be computed as a weighted mean [Huck, 2008]. Based on the World Bank's (2012) definition, the low-income countries in our data set include Nepal, Bangladesh, Cambodia, and Laos. Middle-income countries include Mongolia, Pakistan, Uzbekistan, India, China, the Philippines, Thailand, Vietnam, and Indonesia, while high-income countries include Singapore, S. Korea, Taiwan, and Japan.
 We note in particular the means and standard deviations among countries and within countries between the two time periods. We paid attention to cases when variations are relatively significant, in this case defined as having more than four points difference. This threshold is arbitrary but is nonetheless meaningful. We included in our analysis those cases where there are more than four responses and dropped from the intertemporal analysis cases involving only one respondent and those without comparative data from the 2001/2002 survey period.
 Finally, we interpreted these significant changes from 2001/2002 to 2009/2010 based on objectively verifiable developments in water governance for selected countries, for instance, the introduction of water laws during this period that could have changed perceptions about water governance. We found that most improvements in water governance, say improvements in accountability, greater private sector participation, more integrated approach to water governance, etc. can in fact be explained by the introduction of new water laws, policies, and practices in these countries since 2001.
2.4. Improvements From Saleth and Dinar
 We extend the framework and methods pioneered by Saleth and Dinar in five novel and supplementary ways. First, we collected additional 49 survey responses in 2009∼2010 in addition to the 51 respondents from the Saleth and Dinar survey. This helps to strengthen the robustness of Saleth and Dinar's work. Second, we added three countries—Singapore, Uzbekistan, and Mongolia—which were previously not covered in the Saleth and Dinar survey. Singapore is a useful benchmark country for its best practices in integrated urban water governance.
 Third, we provided an in-depth and nuanced comparison of water governance practices among countries based on income levels. Fourth, we provided an intertemporal analysis within countries for two time periods (2001/2002 and 2009/2010) to understand which aspects of water governance has evolved in some countries in during this period. This paper, therefore, helps to provide a more nuanced (but not perfect and complete) understanding of water governance among countries in Asia and a framework to compare and learn among and within countries overtime.
 There are several limitations to our study. First, it would have been ideal if the size of the survey sample was larger for some countries such as China, India, Indonesia, Bangladesh, and Pakistan. However, because of time and budget constraints, we leave this for future research. Nonetheless, this limitation should have been moderated largely by the reliance on key informants, in this case water expert, to obtain the survey data.
 Second, care has to be taken in interpreting the results for large countries for they are limited to the concerned provinces or states included in the survey rather than for the country as a whole, for instance, China, India, and Pakistan. Future studies would have to collect more data at the provincial or state level to account for the fact that water is often a local good.
 Third, systematic comparison can provide a more nuanced but not perfect picture of the state of water governance among countries in Asia and within countries overtime. Comparative study of water governance across countries and overtime is conceptually and methodologically challenging (but not impossible) to undertake. Fourth, our sample size for high-income countries (Singapore and Japan) is not representative of other high-income countries in Asia (Taiwan and S. Korea). Caution would have to be warranted in their interpretation. Finally, we did not test for the statistical significance of measures of central tendency and variations because of the relatively small size of our sample. This will be left for future research.
3. Results and Discussion
3.1. Overall Finding
 The survey results for the 17 countries for the two time periods (2001/2002 and 2009–2010) are summarized in a comparative table or dashboard (Table 3) for each of the 19 indicators of water governance. We discuss the results in the section that follows.
Table 3. Summary of Findings by Country, Year, and Indicatora
|Mean high income||4.0||State||7.5||7.5||4.0||6.5||BC||4.5||7.5||3.5||3.0||7.5||4.5||7.0||Mixed||8.0||3.0||7.5||6.0|
|Standard dev high income||5.7|| ||0.7||2.8||0.0||0.0|| ||1.4||0.0||0.0||2.8||0.7||2.1||1.4|| ||0.0||7.1||0.7||2.8|
|Mean low income||2.5||State||4.8||5.0||5.0||4.3||BC||4.5||5.0||3.0||4.0||6.3||5.0||5.5||Mixed||5.3||8.3||6.0||4.3|
|Standard dev low income||2.1|| ||0.0||1.4||3.5||0.0|| ||0.7||2.1||1.4||2.8||2.1||0.7||0.7|| ||2.1||4.9||2.1||0.7|
|Mean middle income||6.1||State||4.0||5.0||3.7||3.7||BC||4.6||4.4||2.9||4.3||6.0||3.1||5.7||Mixed||2.6||1.8||5.3||4.9|
|Standard dev middle income||1.5|| ||2.3||2.4||1.9||1.9|| ||1.5||1.8||1.5||2.0||1.9||1.9||1.6|| ||3.2||3.3||1.9||1.8|
3.2. Variations Among Countries
 Table 4 provides comparative summary of various governance indicators amongst the 17 countries covered in the survey. For ease of comparison, we refer back the reader to Table 1 for an operational definition of our variables. We were interested to see how water governance (law, policy, and administration) varies with a country's level of economic development as well as overtime.
Table 4. Summary of Findings (Mean and Standard Deviation Scores) by Level of Economic Developmenta
|Mean high income||2.0||Mult||9.3||8.0||2.0||7.0||BC||3.5||9.0||1.0||1.0||9.3||2.3||8.5||Mixed||10.0||2.5||9.3||7.0|
|Standard dev high income||5.7|| ||0.7||2.8||0.0||0.0|| ||1.4||0.0||0.0||2.8||0.7||2.1||1.4|| ||0.0||7.1||0.7||2.8|
|Mean low income||2.9||Mult||4.7||4.6||5.4||4.3||BC||4.4||5.0||3.1||4.5||6.1||4.9||5.4||Mixed||5.1||7.9||5.8||4.4|
|Standard dev low income||2.1|| ||0.0||1.4||3.5||0.0|| ||0.7||2.1||1.4||2.8||2.1||0.7||0.7|| ||2.1||4.9||2.1||0.7|
|Mean middle income||6.1||Mult||4.3||4.5||4.1||3.9||BC||4.9||4.6||3.2||4.2||6.1||3.5||5.6||Mixed||2.7||2.6||5.5||5.0|
|Standard dev middle income||1.5|| ||2.3||2.4||1.9||1.9|| ||1.5||1.8||1.5||2.0||1.9||1.9||1.6|| ||3.2||3.3||1.9||1.8|
 In summary, our preliminary findings in Table 4 show that, not surprisingly, various aspects of water laws, policies, and administration vary with a country's level of economic development. This result is consistent with Briscoe's  hypothesis about the positive correlation between a country's level of economic development and its state of water governance. This result, if further confirmed by more studies, suggests a similarity to water Kuznet's curve (WKC), i.e., the overall quality of a country's water governance is a function of average income. By implication, as a country's average income increases, its quality of water governance is also expected to increase. As we explain the succeeding sections, this appears to be the case for certain aspects of water law, policy, and administration.
3.2.1. Variations in Water Law
 We find positive correlation between a country's level of economic development and aspects of its water laws, for instance, with (1) legal accountability (L3) for water sector officials (9.3 versus 4.7, 4.3 for high, low, and middle-income countries, respectively); (2) tendency toward centralization (L4) of water governance (8 versus 4.6, 4.5); and (3) more integration of water laws (L6) with other laws on land, forest, and environment (7 versus 4.3 and 3.9). These variations in water laws among high, middle, and low-income countries could simply be the result of more developed legal systems for countries with higher levels of economic development (i.e., spill-over effect).
 In particular, the positive correlation between legal accountability and economic development is consistent with the water governance literature, for instance, Tortajada  in Singapore, Anbarci et al.  in the case of access to drinking water in 85 countries, Davis  in South Asia, Estache and Kouassi , and Plummer and Cross  in Africa. More generally, this is consistent with the empirical literature on governance, i.e., high-income countries also tend to have stronger legal systems, see for instance, the World Bank Governance Index . The findings on greater decentralized water governance in lower income countries is consistent with the literature, for instance [Vermillion, 1997], based on a meta-analyses of 29 irrigation studies.
 We also see a sign of negative correlation between a country's level of economic development and participation (L5) in water governance (5.4 versus 4.2, 2.0 for low, middle, and high-income countries, respectively). This is possibly because of the importance of irrigation (and farmer managed irrigation) in lower income countries [see Lam, 1998; Vermillion, 1997] as well as the importance urban poor water associations in managing water supply in slum areas [McIntosh, 2003].
 However, we find two aspects of water law that do not vary systematically with a country's level of economic development: First, the format of surface water rights (L2) in all countries varied considerably from common or state property to multiple rights, riparian system, appropriative rights, among others, but state property is the most common. This wide variation reflects the unique circumstances that led to the evolution of these rights such as the legal tradition and precedents of a country, its size, geography and water endowments, importance of indigenous water rights, and the country's political system, among other factors. Multiple use water rights is not surprising, for instance, see Bruns et al.  for a more exhaustive conceptual and comparative analysis and Haisman  in Australia.
 Second, the legal distinction (L1) of different water sources (ground, surface, and rain) do not vary systematically with levels of income but perhaps could be better explained by a country's geography, legal origins, and administrative structure. For instance, in middle-income countries (Philippines and Indonesia), variations in the legal distinction of different sources can be explained by variations in the administrative structure of water governance, i.e., there are separate agencies dealing with different types of water sources.
3.2.2. Variations in Water Policies
 In terms of water policy, we find that a country's level of economic development varies with water pricing (P3) with richer countries pursuing more cost recovery (as water tends to be more affordable in these countries); the extent of linkages between water law and policy (P6), and availability of finance for water investments (P8).
 Dinar , based on meta-analyses from 30 countries, concludes that variations in water pricing among countries are largely a function of political economy factors than a country's level of economic development. Lower income countries pay more attention to issues of water and poverty (P7) compared with middle and higher income countries, see, for instance, McIntosh  for a guidebook on improving water access to the urban poor in Asia.
 However, we find little consistent trend among high, middle, and low-income countries in terms of (1) project selection criteria (P1) (i.e., use of benefit-cost analyses, although there are questions if this is actually done in practice) and (2) the extent to which other (nonwater) policies have a significant influence on water policy (P2). This fragmentation is not surprising and is a common critique of scholars of water governance, for instance, Biswas . In the case of water utilities privatization, Clarke et al.  and Hall and Lobina  find mixed results worldwide with more challenging experience from developing countries.
 We also find that a country's level of economic development varies inversely with the extent of private sector participation (P4). It is possible that as a country becomes richer, its public sector becomes equally if not more capable than the private sector in implementing water investment projects. Conversely, in poorer economies, the public sector is relatively weak and hence, may have to rely more on the private sector to implement investment projects.
3.2.3. Water Administration
 Finally, in terms of water administration, we find—not surprisingly—positive correlation between a country's average income and certain aspects of water administration such as (1) functional capacity and balance (A2) among water agencies; (2) use of adequate and reliable water data for planning (A4); and (3) application of science and technology (A5) to solve water governance problems.
 Interestingly, we found that water apex bodies (A3) are more pronounced in low income than high-income countries. There are no clear explanations in the literature for this finding but we speculate that this can be due to the influence of aid agencies in shaping water policies in developing countries. Also, we do not find systematic variations between organizational basis for water (A1) (i.e., geographic, hydrologic, river basin, mixed) and levels of economic development reflecting the unique evolution of institutions in these countries [Saleth and Dinar, 2005; Bruns et al., 2005].
3.3. Variations Within Countries Over Time (2001 and 2010)
 An important contribution of this paper is to explain variations in water governance within countries between 2001 and 2010. We examined in more depth the cases of several countries—the Philippines, Cambodia, Vietnam, Indonesia, and Thailand—to see if there were indeed significant water governance reforms that occurred during this period. We left out in the in-depth analysis countries with single respondents and without follow up survey in 2009/2010. These include Taiwan, S. Korea, Mongolia, Uzbekistan, and New Zealand. We also left out China and India in this analysis because the respondents in both time periods come from different provinces/states and thus, intertemporal comparison is not reliable.
 We find significant changes in many aspects of water governance within countries between these two time periods particularly in Cambodia, Vietnam, and Indonesia but not as much in the case of the Philippines and Thailand. We attribute these significant changes to broader national economic, social and political reforms happening in these countries during this period, which is consistent with the hypothesis of Saleth and Dinar  and Bruns et al. . At the onset, it has to be pointed out that these changes are largely de jure than de facto. Future studies would have to more systematically look at the actual implementation of these laws and what difference would they really make in terms of water sector performance.
 In the case of Cambodia, of the 17 countries covered in the survey, it reported one of most significant changes in water governance since 2001 as a result of the introduction of a comprehensive water resources law in 2007. Changes in water governance indicators that can be attributed directly to the new water resources law includes changes in the legal distinction of different water sources (L1), legal accountability of water sector officials (L3), legal framework for integrated treatment of water sources (L6), linkage between water law and water policy (P6), accountability and regulatory mechanisms (A4), and use of science and technology in water governance (A6). Most of these changes were also the result of success of the internationally acclaimed Phnom Penh Water Supply Authority (PPWSA)—one of the most successful water utilities among developing countries. PPWSA is particularly known for the significant improvements it has introduced in the areas of accountability, integration, and use of science and technology in water governance [see, e.g., Araral, 2008].
 Indonesia, like Cambodia, is 1 of the 17 countries with the most significant changes in water governance since 2001, mainly due to the large scale national reforms introduced in the country during the period of reformasi (1998–2003). These reforms include decentralization to local governments (indicator L4), privatization and liberalization (indicator L5), strengthening of mechanisms of accountability (indicator L3) with the creation of the constitutionally powerful anticorruption agency (Komisi Pemberantasan Korupsi (KPK)), among others.
 In addition, Indonesia likewise adopted a new water law in 2004, which has implications for indicator L1 (legal distinction of different water sources) and L2 (format of surface water property rights), among others. In short, the significant changes in the survey results from 2001 to 2010 in Indonesia can actually be explained by changes in water law, policy, and administration over this period.
 In Vietnam, several indicators of water governance also had significant changes from 2001 to 2010. These include L1 (legal distinction among water sources), L5 (private sector participation), P2 (linkages between water law and policy), P7 (attention to poverty), and P8 (availability of finance for water investments). Again, these changes can be explained by changes in water laws, policies, and administration as a result of the 2006 National Strategy on Water Resources and strengthened by 2005 decrees on river basin and enforcement of water regulations.
 This strategy articulated a number of priority areas for reform, namely (1) national water resources inventory, assessment and water resources database/information system (indicator A5); (2) integrated water resources management (domestic use, irrigation, and hydropower) (indicators L1, P2, and P6); (3) development of inter-reservoirs regulations in important river basins; (4) ground water protection in the major cities; and (5) use of economic instruments on water resources management (P1 and P3). Not surprisingly, therefore, water governance ratings for Vietnam significantly changed in these parameters during this period.
 The Philippines and Thailand also saw significant changes in their water governance since 2001 although not as extensive as Vietnam, Cambodia, and Indonesia. In Thailand, significant changes from 2001 to 2010 were reported in the following indicators: accountability of water sector officials, decentralization, integration, and project selection. Respondents from the 2010 survey in Thailand point to the recent catasthropic flooding of the Chao Phraya River as evidence of the problems of accountability among the provinces in the river basin, inappropriate decentralization of water governance resulting in too little integration of water management at the basin level as well as problematic practices in project selection criteria for water management.
 In the case of the Philippines, out of the 19 water governance indicators, only two indicators had significant changes from the 2002 to 2010 period. The first is the presence of an effective apex of water bodies, in this case the National Water Resources Board, the Local Water Utilities Administration, and the River Basin Control Office at the Department of Environment and Natural Resources, which was a recent creation. The second significant change—a stronger legal distinction of different water sources—is an offshoot of having stronger regulatory water agencies, which are able to enforce regulations related to ground water, surface water, and river basin water.
 This last point is corroborated by a 16-point increase in the score on regulatory accountability (indicator A4). In addition, because of stronger roles for river basin organizations in the country, the rating for decentralization indicator (L4) likewise increased by 28 points after the passage of regulations on river basins. All of these suggest that changes in perceptions on water governance in the Philippines from 2001 to 2010 can be attributed to actual governance changes—more effective water apex bodies and decentralization to river basin organizations.
4. Conclusions and Implications
 We compared water governance practices among and within countries in Asia from 2001 to 2010 by extending the framework and methods pioneered by Saleth and Dinar in five novel and supplementary ways. We surveyed an additional 49 expert respondents in 17 countries and added three new countries. We then provided in-depth and nuanced comparison of water governance practices among countries based on income levels and undertook an intertemporal analysis within countries for two time periods (2001/2002 and 2009/2010).
 There are two main contributions of this paper. First, we have provided a more nuanced (but not a complete and perfect) picture of water governance in 17 countries in Asia. We find that many aspects of water laws, policies, and administration are positively correlated with a country's level of economic development. We find this to be the case in (1) water law (legal accountability for water sector officials, centralization tendency, and integration of water laws with other laws); (2) water policies (water pricing, extent of private sector participation, extent of linkages between water law and policy, and availability of finance for water investments); and (3) water administration (functional capacity and balance among water agencies; use of adequate and reliable water data for planning and application of science and technology). We, however, find a negative relationship between a country's level of economic development and extent of private sector participation in water governance.
 Although these findings are intuitively expected, as far as we know, this is the first systematic comparative study of this kind in the literature. Our initial findings suggest the possibility of water Kuznet's curve, i.e., water governance indicators vary with a country's level of economic development. This result supports Briscoe's  hypothesis about the positive correlation between a country's level of economic development and its state of water governance. However, more studies are needed to confirm our initial findings.
 In contrast, we do not find correlation between a country's level of economic development and several aspects of water governance: legal distinction different water sources, format of surface water rights, project selection criteria, the extent to which other (nonwater) policies have a significant influence on water policy, organizational basis for water and presence of water apex bodies. Because of our small sample size, caution is warranted in making generalizations about the statistical significance from these findings.
 However, the patterns of water governance arrangements that we observed in this study cannot be simply generalized to other countries because governance practices evolve, as we have discussed throughout the paper, according to the unique political, historical, legal, administrative, geographic, and economic circumstances of a country. More studies are needed to make conclusive remarks about the evolution of these patterns of water governance.
 Still, comparison is useful in helping water policy makers learn from and benchmark with the practices of other countries. For instance, Singapore has shown a successful example of integrated water resources management. Manila has shown an example of successful large-scale water utilities privatization and improving service to the urban poor. Phnom Penh has shown best practices in public water utilities and reducing nonrevenue water. China has shown a successful example of integrated river basin management in the Yellow River.
 Second, we have tested and replicated a framework and methodology to compare and learn about water governance within countries overtime. We found significant changes in water governance from 2001 to 2010 in some of countries we surveyed but not in others. We argue that many of these changes can be traced to broader developments in governance in that country—political decentralization, privatization, and liberalization, among others—mostly with donor pressure for reform.
 We now conclude by highlighting several potential prospects for future research in water governance. First, evaluating the impacts of governance reforms is an important area for future research because very few rigorous impact assessment studies exist despite the fact that most scholars agree on its importance. For a start, this study has highlighted several questions for impact assessment. For instance, what difference does it really make to have a more integrated approach to water management? Here, Singapore would be a good case to study. Are there examples of successful and cost effective integrated water management in developing countries?
 Second, what is the impact of having a clearer legal distinction of different water sources or having different formats of water rights? Are there optimal combinations of water rights? Third, what has been the impact of private sector or user participation in terms of water sector performance? Fourth, is decentralization good or bad for water governance? Is it not the case that the privatization of urban water utilities has failed and that irrigation management transfer has produced mixed results? Fifth, what has been the poverty impact of water laws and policies in developing countries? What can we learn from successful examples privatization on one hand and significant improvement in of water service to the urban poor? Sixth, what lessons can be learned from supposedly successful examples of integrated river basin management such as the case of the Yellow River Basin Commission of China? Can these lessons be replicated in other developing countries? Seventh, and finally, which of these governance solutions or bundle of solutions provide the most cost effective means to significantly improve water sector performance? Indeed, a rigorous answer to these questions may lead to a more conclusive answer to how water governance really matters to improving water security in developing countries.
 This project was partly supported by the Institute of Water Policy (2009) and the Staff Research Support Scheme, Lee Kuan Yew School of Public Policy, National University of Singapore.