This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.
Article pubs.acs.org/est
Cultured Construction: Global Evidence of the Impact of National Values on Piped-to-Premises Water Infrastructure Development Jessica A. Kaminsky* Department of Civil and Environmental Engineering, University of Washington 201 More Hall, Seattle, Washington 98195, United States S Supporting Information *
ABSTRACT: In 2016, the global community undertook the Sustainable Development Goals. One of these goals seeks to achieve universal and equitable access to safe and affordable drinking water for all people by the year 2030. In support of this undertaking, this paper seeks to discover the cultural work done by piped water infrastructure across 33 nations with developed and developing economies that have experienced change in the percentage of population served by piped-to-premises water infrastructure at the national level of analysis. To do so, I regressed the 1990−2012 change in piped-to-premises water infrastructure coverage against Hofstede’s cultural dimensions, controlling for per capita GDP, the 1990 baseline level of coverage, percent urban population, overall 1990−2012 change in improved sanitation (all technologies), and per capita freshwater resources. Separate analyses were carried out for the urban, rural, and aggregate national contexts. Hofstede’s dimensions provide a measure of cross-cultural difference; high or low scores are not in any way intended to represent better or worse but rather serve as a quantitative way to compare aggregate preferences for ways of being and doing. High scores in the cultural dimensions of Power Distance, Individualism−Collectivism, and Uncertainty Avoidance explain increased access to piped-to-premises water infrastructure in the rural context. Higher Power Distance and Uncertainty Avoidance scores are also statistically significant for increased coverage in the urban and national aggregate contexts. These results indicate that, as presently conceived, piped-to-premises water infrastructure fits best with spatial contexts that prefer hierarchy and centralized control. Furthermore, water infrastructure is understood to reduce uncertainty regarding the provision of individually valued benefits. The results of this analysis identify global trends that enable engineers and policy makers to design and manage more culturally appropriate and socially sustainable water infrastructure by better fitting technologies to user preferences.
■
INTRODUCTION The newly expired Millennium Development Goals (MDG), a globally agreed upon set of ambitious international development goals, shaped significant global change in recent decades. One of the success stories of the MDG was the provision of improved water infrastructure to 2.6 billion people from the 1990 baseline.1 Indeed, the world met the MDG target for drinking water early, and 91% of the global population now has access to improved sources of drinking water. Still, in 2015 there were 663 million people worldwide without access to an improved source of water,2 which is internationally defined as access to piped water to a dwelling or plot, rainwater, a public tap or standpipe, a tubewell or borehole, a protected dug well, or a protected spring.2 In this framework, these technologies are assumed to provide reasonable levels of protection from contaminated water and serve as proxies for water quality standards. Just as troubling as the count of people still without coverage are the social inequities reflected in the demographics of the people who compose that 663 million. Populations without access to improved water technologies are more rural than © 2016 American Chemical Society
urban and more poor than rich, and the burden of water collection (long walks with heavy loads resulting in missed school and physical strain) falls most heavily on females and children.3,4 Moving from demographics to geography, a pipedto-premises water supply serves nearly 80% of urban populations but just 33% of rural populations.5 However, percent coverage in urban areas has barely changed over the tenure of the MDG, while the rural context has made substantial progress; in fact, in sub-Saharan Africa urban coverage expressed as a percentage of population has actually declined over the tenure of the MDG.5 While this is partially due to rapid increases in urban populations, it also highlights the difficulty of reaching all people, especially those living in poverty. Reflecting the work yet to be done, water infrastructure stands large in the post-2015 global development agenda,6 Received: Revised: Accepted: Published: 7723
March 3, 2016 May 20, 2016 June 7, 2016 June 7, 2016 DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology which by 2030 aspires to “achieve universal and equitable access to safe and affordable drinking water for all.” Indicators for the new Sustainable Development Goals (SDG) will measure safely managed drinking water provision,6 currently defined as “a basic drinking water source which is located on premises, available when needed and free of faecal and priority chemical contamination.”7 This explicit focus on water quality is a significant policy change from the improved water technologies previously measured in the MDG. At the global aggregate, the most frequently adopted technology for household water supply is water that is piped directly to premises; in 2015, this technology served 58% of the global population.3 As detailed below, although the prevalence of this technology varies by geography, 1.9 billion of the people who gained access to an improved drinking water source during the tenure of the MDG gained access via pipes.1 In addition, the recent shift to the SDG definitions likely means that a piped-to-premises supply will continue to be a preferred water infrastructure technology. For example, recent research8 shows that on-plot water supply is a useful proxy for human health outcomes and thus is a useful benchmark for measuring progress toward the SDG. Given the evidence that a piped-topremises water supply results in improved human health outcomes, it is therefore important for engineers and policy makers to discover diverse reasons why people choose this versus other technologies such as well water. There is good reason to believe that culture influences technology choice for water infrastructure; for example, while in Eastern Asia, Latin America, and the Caribbean, progress in access to improved water sources has been driven by piped water, coverage in Southern Asia and sub-Saharan Africa have been dominated through other improved technologies, such as protected springs and wells.5 This observed geographic variation begs an explanation; after all, if infrastructure technology were perfectly value-neutral, we might expect variation in the construction of piped-to-premises water infrastructure to be perfectly explained by a combination of access to economic and water resources and population density (all of which are controlled for in this study). Furthermore, it is known that cultural tendencies impact technology choice in other types of civil infrastructure. For example, Hofstede’s cultural dimensions have been shown to have statistically significant relationships at the national level with technology choice for both sanitation and electrical infrastructure systems.9,10 However, we lack empirical research that explains how Hofstede’s cultural dimensions relate to water infrastructure, which (practically speaking) means we are less able to create culturally appropriate water infrastructure systems. In addition, the bulk of the evidence regarding the importance of considering culture in infrastructure is smaller-scale research that considers, at most, tens of cases. In contrast, the work presented here seeks global trends across 33 nations with both developed and developing economies that have experienced growth in the percentage of population served by piped-topremises water infrastructure between 1990 and 2012. The overarching exploratory research question for this paper is How do Hofstede’s cultural dimensions11 explain national variation in change in access to a piped-to-premises water supply over the tenure of the Millennium Development Goals? It is worth re-emphasizing that the analysis and theory presented here apply to both developed and developing economies.12 The fundamental theoretical perspective underlying this research is that if cultural preferences can explain variation in the
construction of water infrastructure, then that infrastructure is doing some form of culturally relevant work. For example, and as will be shown in this paper, as it is presently realized, piped-topremises water infrastructure appears to be better suited to contexts that prefer organizational hierarchy (higher Power Distance dimension per Table 2). There are certainly other water infrastructure technologies that may be a better fit for contexts that eschew organizational hierarchies; similarly, one can imagine less hierarchical ways of organizing a piped-topremises water supply. These are places for future research and policy attention. By making cultural work explicit, we can learn how to create more culturally appropriate and socially sustainable designs and policy. This will enable us to change technology to better serve people. Hofstede’s Cultural Dimensions. To use national values to predict variation in the construction of piped-to-premises water infrastructure, I need a widely available, quantitative metric for comparative national values. One such metric is Hofstede’s cultural dimensions.11 Ending in the early 1970s, Hofstede administered nearly 90 000 surveys to employees of IBM across the globe. He used the results of this survey to develop four cultural dimensions that are intended to provide comparative cross-cultural information. For example, one of these dimensions is Individualism−Collectivism. The United States is the most highly individualistic nation measured, with a score of 91 for this dimension. In contrast, Ecuador has the second lowest score of any nation on this dimension, with a score of 8. These scores represent placement on a descriptive continuum and are not intended to imply positive or negative cultural attributes but rather represent distance between nations as a measure of cross-cultural difference. In addition, nations that score differently on one dimension may be quite similar in another dimension. For example, the United States and Ecuador score almost identically on Hofstede’s Masculinity− Femininity dimension (62 and 63, respectively). The Masculinity-Femininity (MAS) dimension is defined as “a preference in society for achievement, heroism, assertiveness, and material rewards for success. Society at large is more competitive” versus a “preference for cooperation, modesty, caring for the weak and quality of life. Society at large is more consensus-oriented.”13 The MAS dimension has strong descriptive power for cultural tendencies toward gender roles and work behaviors; however, I do not discuss these here. MAS scores have been linked to technology through studies of consumer preferences.14,15 For example, de Mooji links low MAS scores to maker cultures in which the homemade (dresses, cigarettes, and home repairs) are preferred. De Mooji also identifies a correlation between status purchases and high MAS scores (watches and car engine power). Of particular relevance to this study is the fact that low MAS scores have also been linked to increased aid for the poor.11 Similarly, although there is debate in the literature,16,17 low MAS scores are often correlated with environmental concerns,18−20 and people in nations with low MAS scores are more easily mobilized for social issues such as the environment.11,21 The Individualism-Collectivism (IDV) dimension is defined as “a preference for a loosely-knit social framework in which individuals are expected to take care of only themselves and their immediate families“ versus a “preference for a tightly-knit framework in society in which individuals can expect their relatives or members of a particular in-group to look after them in exchange for unquestioning loyalty.”13 In other words, this dimension describes the relationship between the individual 7724
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology
Figure 1. Nations included in the analysis.
practices may be ritual; they may not actually change outcomes.30 Regardless, they enable members of a society to tolerate uncertainty. High UAI scores also indicate preferences for larger organizational size11 and more structured activity.31 Finally, there is some evidence that uncertainty-accepting nations (low UAI) tend to be more accepting of innovation32−35 than are their high UAI peers. Like any framework, there are serious limitations of Hofstede’s cultural dimensions.36−40 An important methodological criticism is the use of nation states as a unit of analysis; nations are certainly not culturally homogeneous. However, infrastructure policy is often set at the national level, making this an important scale for the current analysis. In another limitation, Hofstede has declined to update his dimensions, which were originally developed in the early 1970s. A particularly key criticism is that culture is a complex topic that cannot reasonably be represented with a handful of numbers. Still, however limited Hofstede’s cultural dimensions may be, my purpose here is to determine if they have predictive power that can explain variation in the construction of pipedto-premises water infrastructure. The value of this undertaking is as a global screening analysis to identify cultural tendencies that are likely to be analytically valuable for more detailed qualitative studies that can do a better job of operationalizing and problematizing the cultural tendencies that emerge as significant here. In combination with this future work, this may permit us to build more culturally appropriate and thus more socially sustainable water infrastructure for all of the world’s population as we pursue the SDG.
and the social collective(s) they live in. For example, the IDV dimension has descriptive power for how individuals confirm to group behaviors, 22 such as new technology adoption. Accordingly, the IDV dimension has also been linked to the nature of the relationship between individuals and organizations and the nature of the technology developed in those organizations.23 The bias of particular technologies toward the individual or collective has been previously identified as an issue for international technology diffusion and movements for appropriate technology.24 In a further implication for the diffusion of technology, high IDV has been linked to higher levels of innovation at the national scale.25 Other research has used IDV to explain variation in innovation strategies; for example, innovators in nations with low IDV scores tend not to innovate alone but rather to involve others.26 The Power Distance (PDI) dimension is defined as “the degree to which the less powerful members of a society accept and expect that power is distributed unequally. The fundamental issue here is how a society handles inequalities among people.”13 Nations with high PDI tend to prefer vertical hierarchies in organizations, and it is more acceptable for individuals with power to have privileges that others do not. For example, high PDI has been correlated to lower charitable behavior,27 and we might thus expect a high PDI to lead to lower rates of access to water infrastructure for the poor. In contrast, as high PDI is correlated with concentration of power and decision making,28 we might instead expect that in high PDI nations, centrally managed networked infrastructure would be more culturally appropriate and thus more ubiquitous. The Uncertainty Avoidance (UAI) dimension “expresses the degree to which the members of a society feel uncomfortable with uncertainty and ambiguity.”13 High UAI indicates that a nation is uncomfortable with uncertainty; low UAI nations are more tolerant of uncertainty. UAI has a strong link to technology, as this is one of the strategies used to manage or at least structure unavoidable uncertainty.11 UAI also has a strong link to organizations; indeed, Hofstede borrowed the term from foundational organizational theory work.29 In this theoretical tradition, technology and organizational rules are used to increase predictability. It is worth noting that these
■
MATERIALS AND METHODS The data for this research originates from online databases detailed in the Supporting Information. Figures for piped-topremises water coverage and overall improved water infrastructure coverage from 1990 and 2012 are published by the World Health Organization−UNICEF Joint Monitoring Program’s Water Supply and Sanitation database.3 Hofstede’s cultural dimensions are published on Hofstede’s Web site.13 I caution that the Hofstede scores are biased to the urban context because they were created from employees of IBM offices. 7725
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
7726
a
11586 30 22
9921 62
79
7.3 7.8
9.3
9.2
14675
17 19 19 23
48 34 66 73
12463
14
24
31
6
8
43
9
std deviation
15
mean
*, p < 0.05; **, p < 0.01; ***, p < 0.001.
outcomes 1. percent Δ national piped, 1990−2012 2. percent Δ urban piped, 1990−2012 3. percent Δ rural piped, 1990−2012 predictors 4. MAS 5. IDV 6. PDI 7. UAI controls 8. GDP per capita, 2012 9. percent national baseline piped water, 1990 10. percent urban baseline piped water, 1990 11. percent rural baseline piped water, 1990 12. m3 freshwater per capita, 2012 13. percent urban population change 1990−2012 14. percent improved water infrastructure change 1990−2012
measure
−0.49** −0.50** −0.46** −0.51** −0.10
−0.57*** −0.44* −0.19 −0.53** −0.05
0.54**
0.52**
0.08
−0.07 −0.41* 0.44* −0.16
−0.1 −0.48** 0.39* −0.13
0.39*
0.14
1
2
0.50**
0.68***
1
1
0.05
0.03
−0.11 0.02
−0.31
0.06
−0.08
0.19 −0.04
1 0.22 0.12 −0.15
4
0.33
−0.14
0.36*
−0.33 0.22
−0.22 −0.10 0.05 0.31
1
3
0.22 0.48**
−0.46**
0.09
−0.44*
−0.39*
−0.58*** −0.42*
1 −0.08
6
−0.53**
−0.25
0.47**
0.41*
0.60*** 0.47**
1 −0.59*** −0.10
5
−0.49**
−0.15
0.21
0.38*
0.50**
0.06 0.53**
1
7
−0.58***
−0.18
−0.03
0.77***
0.53**
1 0.64***
8
−0.75***
−0.26
0.28
0.90***
0.94***
1
9
Table 1. First-Order Correlations between Change in Piped-to-Premises Water Infrastructure and Hofstede’s Cultural Dimensionsa
−0.63***
−0.15
0.27
0.80***
1
10
−0.71***
−0.14
0.04
1
11
−0.10
−0.03
1
12
0.40*
1
13
1
14
Environmental Science & Technology Article
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology Table 2. Piped Water Regression Results
a
pa
drives more piped water
0.8 0.7 125.9 51.7 546.2 608.0 4.0 42.2 31.3
0.44 0.49 0.000*** 0.000*** 0.000*** 0.000 0.001* 0.000*** 0.000***
− − higher PDI higher UAI higher GDP more water lower start more urban migration more water infrastructure
2.2 0.1 0.1 0.0 0.0 0.0 0.4 0.2 0.1
0.4 7.1 7.3 115.3 387.1 84896.3 2.7 2.4 7.9
0.67 0.000*** 0.000*** 0.000*** 0.000*** 0.000*** 0.01** 0.02* 0.000***
− higher IDV higher PDI higher UAI higher GDP more water lower start less urban migration more water infrastructure
0.8 0.7 0.0 0.0 0.0 0.0 0.3 0.2 0.0
1.3 1.3 217.1 92.8 679.3 2888.8 3.8 4.0 62.9
0.23 0.22 0.000*** 0.000*** 0.000*** 0.000*** 0.001** 0.001** 0.000***
− − higher PDI higher UAI higher GDP more water lower start less urban migration more water infrastructure
b
(SE)
national piped water change 1990−2012 R2 = 0.35 MAS (b1) IDV (b2) PDI (b3) UAI (b4) GDP per capita, 2012 (b5) freshwater per capita, 2012 (b6) national piped water, 1990 (b7) urban population change 1990−2012 (b8) Δnational improved water coverage, 1990−2012 (b9)
0.98 0.95 1.02 1.01 1.00 1.00 0.99 1.11 1.07
1.3 1.3 0.0 0.0 0.0 0.0 0.2 0.0 0.0
rural piped water change 1990−2012 R2 = 0.20 MAS (b1) IDV (b2) PDI (b3) UAI (b4) GDP per capita, 2012 (b5) freshwater per capita, 2012 (b6) rural piped water, 1990 (b7) urban population change 1990−2012 (b8) Δnational improved water coverage, 1990−2012 (b9)
0.98 1.01 1.00 1.02 1.00 1.00 0.99 0.57 1.04
urban piped water change 1990−2012 R2 = 0.39 MAS (b1) IDV (b2) PDI (b3) UAI (b4) GDP per capita, 2012 (b5) freshwater per capita, 2012 (b6) urban piped water, 1990 (b7) urban population change 1990−2012 (b8) Δnational improved water coverage, 1990−2012 (b9)
0.98 0.93 1.02 1.01 1.00 1.00 0.99 0.65 1.08
t
*, p < 0.05; **, p < 0.01; ***, p < 0.001. Nonlinear least-squares analysis. All coefficients initialized at 1. n = 33 nations.
Finally, the remaining control data I considered (GDP per capita, water resources per capita, and change in percent urban population) were taken from the World Bank’s World Development Indicators Database.41 These control variables allow us to statistically isolate how much each control variable and each cultural dimension explains variability in the construction of piped-to-premises water infrastructure. Data from these databases were matched to eliminate nations with missing data values. The data were filtered to eliminate nations that had not experienced growth in the coverage of urban water infrastructure (the most restrictive context). This resulted in 33 nations being included in the analysis. Figure 1 shows a map of these nations, which are also listed in the Supporting Information for this article. An immediately apparent data limitation of particular importance to an analysis concerned with increasing access to water infrastructure is how few nations from the African continent are included. This is due to the small number of African nations with Hofstede measures available. Using STATA 14.1, I checked the residuals to ensure assumptions for linearity, normality, and homoscedasticity were met. The Shapiro−Wilk test for normality of residuals supports the normality of the data, although graphs of the residuals
showed a slight skew. However, there were significant nonlinearities in the predictor variables, and various transformations did not produce acceptable results. Correspondingly, the largest mean variance inflation factor was 2.38, well below the threshold of concern for multicollinearity. To investigate these nonlinearities further, I checked for outlying data points that might bias the results. A Cooks distance estimate shows no severe outliers, but Guatemala appears as a high leverage point. As a further check for outliers, I used the graphical tool proposed by Rousseeuw and van Zomeren42 using the procedure presented in Verardi and Croux.43 This procedure identified an outlying cluster of Asian nations (Indonesia, Bangladesh, India, Vietnam, Philippines, Pakistan, China, Thailand, and Malaysia). However, because this cluster represented a significant portion of the data set, it was not appropriate to exclude these as outliers. First-order correlations (Table 1) indicated the presence of statistically significant relationships between the cultural dimensions, control variables, and outcomes of interest. Of particular note is that PDI and IDV are significantly correlated with each other. Leave-one-out cross-validation did not show a strong preference for retaining one of these and not the other in terms of reducing root mean 7727
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology
is a fundamental requirement of human life. However, this is not what the data indicate in either first-order correlations (Table 1) or the exponential regression model (Table 2, rural, urban and national PDI, b ≥ 1.0). I explain the relationship between PDI and increased pipedto-premises water coverage with reference to organizational size. Piped water networks are hierarchical, complex systems; this type of organization is known to correlate to high PDI scores.28 The relationships discovered here suggest that, as currently enacted, piped water infrastructure is a better cultural fit in settings that tend toward more centralized decision structures, vertical hierarchies, and formal management rules. This in turn suggests the need for future research and policy to support the development of alternate and less-centralized structures for both technical and managerial aspects of water infrastructure for use in low PDI contexts. Once again, qualitative research is needed to validate these explanations of the observed relationships. Uncertainty Avoidance. For the rural, urban, and national aggregate contexts, more discomfort with uncertainty (high Uncertainty Avoidance [UAI] scores) correlate with a higher increase in the percentage of national populations with piped water on premises (Table 2, UAI, b > 1.0 and p < 0.05). In the rural context, UAI is the statistically dominant variable (Table 2, b = 1.02). High UAI scores mean that nations are less comfortable with uncertainty; one strategy for dealing with this uncertainty is to structure it through the use of technology.11 Although the current analysis cannot tell us what types of uncertainty are being managed by water infrastructure, we might suppose they include the risk of inadequate water quantity or quality, risk of illness due to contaminated water, or risk regarding the time required to gather and transport water. Future qualitative research is needed to empirically discover these factors, noting that they are likely context dependent. Individualism−Collectivism. For the rural context only, highly individualistic nations (high Individualism−Collectivism [IDV] scores) correlate with a higher increase in the percentage of national populations with piped water on premises (Table 2 rural IDV, b > 1.0 and p < 0.05). In urban environments, there is not a statistically significant relationship (Table 2, urban IDV, p > 0.05), and this combination translates to the absence of a significant relationship for the national aggregate (Table 2, national IDV, p > 0.05). High IDV indicates preferences toward the individual rather than toward collective. The (rural context) correlation between high IDV and increased water piped-topremises may seem counterintuitive to engineers who see the piped water infrastructure as a collective network. However, it would seem that at the user level, this systems level perception is not the norm. Instead, in the rural context, piped-to-premises water infrastructure is seen as an individual benefit to the individual family. Why, then, do we not observe significant relationships between IDV and water infrastructure in the urban context? A possible explanation is that denser urban environments showcase the collective nature of water infrastructure, as in urban environments piped-to-premises water infrastructure tends to be part of a larger system than is common in rural contexts. This emphasizes the collective nature of centralized infrastructure, and this may culturally interfere with the individual appeal of a household connection. Alternatively, in a rural environment, an individual might (individually) lead construction of a new supply system, whereas in an urban environment, connecting to an existing (collective) system is
squared error; as such, I retained both in the regression model presented here. As a result of these combined analyses, and drawing from past work on cultural dimensions and infrastructure technology choice,10 I undertook a nonlinear least-squares analysis that tested the fit of common logistic, Gompertz, and exponential functions to the data. The exponential growth model provided the best fit and was used for the results presented here; eq 1 below shows the regression equation. Δpiped water = b0 + b1MAS + b2 IDV + b3PDI + b4 UAI + b5GDP + b6 Water + b7 Baseline + b8ΔUrban + b9ΔImproved
(1)
where Δpiped water is the 1990−2012 change in the percent of national population served by piped-to-premises water infrastructure in the urban, rural, and national aggregate contexts; bn are constants; IDV is the Individualism−Collectivism dimension; PDI is the Power Distance dimension; UAI is the Uncertainty Avoidance dimension; GDP is per capita 2012 gross domestic product; Water is per capita 2012 freshwater resources; Baseline is the 1990 baseline coverage of piped-topremises water expressed as a percent of the national population and specific to the urban, rural, or national aggregate context; ΔUrban is the change in percent urban population between 1990 and 2012, and ΔImproved is the change in percent improved water infrastructure between 1990 and 2012.
■
RESULTS AND DISCUSSION The regression results (Table 2) indicate that piped-topremises water infrastructure is culturally understood as an individual level benefit (high IDV) that reduces uncertainty (high UAI). As currently designed and operated, it fits best with nations that prefer hierarchy and centralized control (high PDI). These relationships are robust after controlling for per capita GDP, per capita freshwater resources, the coverage of piped water infrastructure at the 1990 baseline, the change in percent urban population, and the change in overall improved water infrastructure coverage. These various claims are elaborated below in the sections indicated by the above parenthetical text, with suggestions for future qualitative research that can validate the proposed explanations of the observed statistical relationships. The results of this analysis are characteristics of piped-to-premises water infrastructure that policy makers, engineers, and community members should be aware of as they select technologies and features for water supply. By fitting water infrastructure technology to local cultural context, we can better pursue socially sustainable water infrastructure. This in turn supports infrastructure longevity, the SDG, and public health goals. Power Distance. In the national and urban contexts, the Power Distance Index (PDI) is the statistically dominant factor for the construction of piped-to-premises water supply between 1990 and 2012; this can be seen in Table 2 by noting that PDI has the highest coefficient (1.02). At the urban, rural, and national aggregate levels, high comfort with inequality (high PDI) correlates to differentially increased piped-to-premises water construction. We might have hypothesized that poor segments of the population would have gained differentially more access to piped-to-premises water in contexts that are less comfortable with inequalities (or with low PDI). After all, water 7728
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology
structure construction. The results support this hypothesis (Table 2, urban population change, p < 0.000). The final control variable considered the overall change in improved water infrastructure with all technologies combined. This significant control variable (Table 2, Δnational improved water coverage, p < 0.000) is a proxy for a host of other contributing variables such as good governance or external development aid that influence the construction of improved sanitation. This control variable concentrates the analysis on water infrastructure technology choice rather than the higherlevel decision of whether or not to build water infrastructure at all. Each of the control variables considered here can be considered as an alternative hypothesis to my claim that cross-national cultural difference explains some of the variation in the construction of piped-to-premises water infrastructure technologies. However, after these variables were controlled for, several of the cultural dimensions are still statistically significant across the various contexts (Table 2). National, Rural, and Urban Contexts. Although the above discussion treats each dimension individually, it has not yet considered what the combined results mean for particular contexts. The analysis presented here is yet another reminder of the importance of disaggregating rural and urban populations for infrastructure analyses. For example, Table 2 shows that the rural context has three statistically significant cultural dimensions (IDV, PDI, and UAI), while urban context has two (PDI and UAI). An explanation for this is that the density of the urban environment physically or institutionally46 constrains networked infrastructure sufficiently to obviate weaker cultural influences; we can see this in Table 2 by noting that the coefficients for PDI and UAI are higher than those for MAS or IDV. Institutions are certainly culturally influenced,46 and the data reflect this. However, it seems reasonable that an urban environment would be more technically formalized, borrowing from other (foreign) urban contexts with developed systems and leading to a more homogenized technical and organizational outcomes.47 To validate this, future research could discover if variation in regulatory water policies is explained by the PDI and UAI dimensions. In other words, these dimensions may remain significant in the more highly regulated urban environment because they influence those regulations more directly than do MAS or IDV. In contrast, rural systems appear to have more regulatory flexibility and thus are more deeply shaped by local cultural factors. Policy Implications. Across the tenure of the MDG, Hofstede’s cultural dimensions explain variation in the increased coverage of piped-to-premises water infrastructure across 33 nations. In the rural context, a higher change in coverage corresponds to higher UAI, IDV, and PDI scores. In the urban context, higher change in coverage corresponds to higher PDI and UAI scores. At the national aggregate, both PDI and UAI are significant. These relationships between water infrastructure and Hofstede’s cultural dimensions are robust to control for the 1990 baseline coverage of piped water infrastructure, per capita GDP, per capita national water resources, change in percent urban population, and change in overall improved water infrastructure coverage. I claim that this discovery can help make water infrastructure design and policy more culturally appropriate. I also claim that the factors proposed in this paper are broadly applicable. For example, although some nations may be more culturally comfortable
more likely. This latter situation is likely independent of the urban−rural setting and may instead reflect the phase of infrastructure development. In other words, this may reflect the preferences regarding connecting to existing infrastructure and constructing new infrastructure rather than to urban−rural differences. However, without infrastructure system level data, we cannot verify this. Once again, additional research is needed to validate these hypothesized explanations. Masculinity−Femininity. The Masculinity−Femininity (MAS) dimension was not statistically significant for any of the contexts considered in this analysis (Table 2, urban, rural, national MAS, p > 0.05). In terms of consumer behavior, the MAS dimension is related to status purchases.15 Status is differentiating and creates a ranking of those who have and those who have not;44 water infrastructure seems to transcend this category. Instead, low MAS is known to be related to environmentalism. The lack of a significant relationship with MAS indicates that an environmental framing does not significantly explain variation in the construction of piped-topremises water infrastructure. This may be because the data measures increases in water supply rather than improvements in water quality or because piped water is understood as a consumable component of broader, environmental water resources. Future qualitative research is needed to explore these proposed explanations of the observed trends. Control Variables. Unsurprisingly, higher GDP per capita leads to a higher increase in coverage of piped-to-premises water (Table 2, urban, rural, national GDP, b > 1.0 and p < 0.05). Similarly, the coverage of piped water infrastructure at the 1990 baseline is also significant in our model (Table 2, urban, rural, national piped water baseline, p < 0.05). Nations with lower water coverage at the baseline built differentially more piped water infrastructure (Table 2, urban, rural, national piped water baselines, p < 0.05 and b < 1.0). This is true in both wealthy and poor nations, and in both urban and rural contexts, despite the differing extents of infrastructure coverage in these various contexts. National per capita water resources were also statistically significant in the model. Higher freshwater resources per capita predicts differentially increased access to piped-to-premises water coverage (Table 2, urban, rural, national piped water baselines, b > 1.0 and p < 0.05). Interestingly, per capita freshwater resources are not significantly correlated with the 1990−2012 change in overall improved water infrastructure (Table 1). In other words, more water resources do not imply more improved water infrastructure overall but rather is specifically related to the construction of piped-to-premises supply. It is possible the observed relationship between freshwater resources and the construction of pipes happens because contexts with the most water have access to larger, centralized supplies (for example, water from a large river or aquifer) that can be efficiently input into piped networks, while contexts with less water need to access smaller, more distributed sources (for example, household scale wells or rainwater collection). It is worth adding the caveat that nations with absolute per capita water scarcity are not well represented in our data set because water scarce nations are typically African nations45 that, as noted earlier, were not scored by Hofstede. The inclusion of the percent urban population control variable is important for water infrastructure given the sheer number of urban migrants in recent decades and the astounding growth of peri-urban slums; in other words, change in the population variable would reasonably be expected to explain variation in water infra7729
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology with uncertainty than others, I would suspect that virtually all of them prefer less uncertainty regarding water supply. I interpret these statistical results as suggesting that piped-topremises water infrastructure is a complex system that, as currently realized, culturally fits best with collectives that prefer hierarchy and centralized decision-making (high PDI). The benefits resulting from these complex systems are valued at the individual level (high IDV). A key benefit provided by piped-topremises water infrastructure is the reduction of uncertainties (high UAI) related to water supply. Additional qualitative research is needed to validate these explanations of the observed statistical relationships. Particularly fruitful aspects to study include the types of uncertainty that water infrastructure is perceived to reduce and research on more organizationally horizontal designs, policy, and management approaches for piped-to-premises water infrastructure. In addition, as infrastructure projects are undertaken at the local level, future research at this unit of analysis is needed. Similarly, additional research is needed to explore connections between the urban versus rural contexts and the ways people are aware (or not48) of their water infrastructure. Due to data limitations, African nations are not well represented in the current analysis. Given the relatively high number of people on the African continent without access to improved drinking water, research is urgently needed to contextualize and validate these findings for African nations. Finally, for policy makers and development workers interested in water infrastructure, the data recommend strategies that target individual benefits and that match the horizontal or vertical organization for the design and management of water infrastructure systems to local preferences. As the new SDG goals are written into policy, nations with both developed and developing economies will be measured against these new metrics. The new SDG set a higher bar for public health protection through water infrastructure than did the now-expired MDG indicators. Accordingly, many nations will fall short in 2016; the more interesting question is how many will do so by the target date of 2030. Recent evidence shows that on-plot water supply tends to lead to improved health outcomes and may be appropriate as an indicator of the safe water required by the SDG.8 This makes piped-to-premises supply a particularly important technology for the coming decades. The evidence presented here shows that cultural trends, not just technical or economic forces, influence the construction of piped-to-premises water infrastructure as opposed to other water infrastructure technologies. This paper describes one model for describing those cultural trends that is statistically significant across the globe and makes some specific recommendations for ways to consider those global trends in infrastructure policy and organization. The most important contribution of this paper is more general: it provides statistical evidence that, if we are to reach the SDG, the social pillar of sustainability must be considered in water infrastructure policies. This in turn is a call to action across policy and research communities that must now discover systematic ways to localize these findings and change the design and regulation of water infrastructure accordingly.
■
■
Tables showing water infrastructure metrics and Hofstede’s cultural dimensions. (PDF)
AUTHOR INFORMATION
Corresponding Author
*Telephone: +1-206-543-1543; fax: +1-206-221-3058; e-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS I thank the University of Washington for financial support of this work.
■
REFERENCES
(1) UN. The Millennium Development Goals Report; United Nations: New York, NY, 2015; p 75. (2) WHO/UNICEF JMP. WHO/UNICEF Joint Monitoring Programme: wat/san categories. http://www.wssinfo.org/definitionsmethods/watsan-categories/ (accessed Aug 24, 2015). (3) WHO/UNICEF JMP. WHO/UNICEF Joint Monitoring Programme: tables. http://www.wssinfo.org/data-estimates/tables/ (accessed Aug 24, 2015). (4) Ray, I. Women, water, and development. Annu. Rev. Environ. Resour. 2007, 32, 421−449. (5) WHO/UNICEF JMP. Progress on sanitation and drinking water: 2015 Update and MDG Assessment; WHO/UNICEF Joint Water Supply and Sanitation Monitoring Programme; WHO/UNICEF: Geneva, Switzerland, 2015. (6) UN Statistical Comission. Report of the Inter-Agency and Expert Group on Sustainable Development Goal Indicators; United Nations: New York, NY, 2016. (7) UN-Water. UNSSO Statement Goal 6; IAEG-SDG: Bangkok, 2015. (8) Overbo, A.; Williams, A. R.; Evans, B.; Hunter, P. R.; Bartram, J. On-plot drinking water supplies and health: A systematic review. Int. J. Hyg. Environ. Health. 2016, 219, 317. 10.1016/j.ijheh.2016.04.008 (9) Kaminsky, J. Cultured Construction: Global Evidence of the Impact of National Values on Sanitation Infrastructure Choice. Environ. Sci. Technol. 2015, 49 (12), 7134−7141. (10) Kaminsky, J. Cultured Construction: Global Evidence of the Impact of National Values on Renewable Electricity Infrastructure Choice. Environ. Sci. Technol. 2016, 50 (4), 2108−2116. (11) Hofstede, G. Culture’s consequences: comparing values, behaviors, institutions, and organizations across nations; Sage Publications: Thousand Oaks, CA, 2001. (12) UN. Statistical Annex: Country Classification. http://www.un. org/en/development/desa/policy/wesp/wesp_current/2012country_ class.pdf (accessed Nov 18, 2014). (13) Hofstede, G. Dimensions - Geert Hofstede. http://geerthofstede.com/dimensions.html (accessed Feb 18, 2014). (14) De Mooij, M. Global Marketing and Advertising: Understanding Cultural Paradoxes: Understanding Cultural Paradoxes; Sage Publications: New York, 2013. (15) de Mooij, M. Masculinity/Femininity and Consumer Behavior. In Masculinity and Femininity: The Taboo Dimension of National Cultures; SAGE: Thousand Oaks, CA, 1998. (16) Onel, N.; Mukherjee, A. The effects of national culture and human development on environmental health. Environ. Dev. Sustain. 2014, 16 (1), 79−101. (17) Power, D.; Klassen, R.; Kull, T. J.; Simpson, D. Competitive Goals and Plant Investment in Environment and Safety Practices: Moderating Effect of National Culture. Decis. Sci. 2015, 46 (1), 63− 100. (18) Dunlap, R. E.; Gallup, H. G.; Gallup, A. M. Of Global Concern. Environment 1993, 35 (9), 7−39.
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.6b01107. 7730
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
Article
Environmental Science & Technology
(43) Verardi, V.; Croux, C. Robust regression in Stata. FBE Res. Rep. KBI0823 2008, 1−13. (44) Deephouse, D. L.; Suchman, M. C. Legitimacy in Organizational Institutionalism. In The SAGE Handbook of Organizational Institutionalism; Greenwood, R., Oliver, C., Sahlin-Andersson, K., Suddaby, R., Eds.; SAGE: New York, 2008. (45) UN WATER. Water for a sustainable world: World Water Report; Unesco: Paris, 2015. (46) Scott, W. R. Institutions and Organizations: Ideas, Interests, and Identities, 4th ed.; SAGE Publications, Inc: Los Angeles, 2013. (47) Meyer, J. W.; Rowan, B. Institutionalized Organizations: Formal Structure as Myth and Ceremony. Am. J. Sociol. 1977, 83 (2), 340− 363. (48) Star, S. L.; Ruhleder, K. Steps toward an ecology of infrastructure: Design and access for large information spaces. Inf. Syst. Res. 1996, 7 (1), 111−134.
(19) Park, H.; Russell, C.; Lee, J. National culture and environmental sustainability: A cross-national analysis. J. Econ. Finance 2007, 31 (1), 104−121. (20) Peng, Y.-S.; Lin, S.-S. National culture, economic development, population growth and environmental performance: The mediating role of education. J. Bus. Ethics 2009, 90 (2), 203−219. (21) WVS. WVS Database. http://www.worldvaluessurvey.org/ WVSContents.jsp (accessed Jan 29, 2015). (22) Bond, R.; Smith, P. B. Culture and conformity: A meta-analysis of studies using Asch’s (1952b, 1956) line judgment task. Psychol. Bull. 1996, 119 (1), 111. (23) Stinchcombe, A. Social structure and organizations1. Handb. Organ. J. March Ed. 1965, 1425193. (24) Cerne, M.; Jaklic, M.; Skerlavaj, M. Decoupling management and technological innovations: Resolving the individualism-collectivism controversy. J. Int. Manag. 2013, 19 (2), 103−117. (25) Taylor, M. Z.; Wilson, S. Does culture still matter?: The effects of individualism on national innovation rates. J. Bus. Ventur. 2012, 27 (2), 234−247. (26) Shane, S.; Venkataraman, S.; MacMillan, I. Cultural differences in innovation championing strategies. J. Manag. 1995, 21 (5), 931− 952. (27) Winterich, K. P.; Zhang, Y. Accepting Inequality Deters Responsibility: How Power Distance Decreases Charitable Behavior. J. Consum. Res. 2014, 41 (2), 274−293. (28) Daniels, M. A.; Greguras, G. J. Exploring the Nature of Power Distance Implications for Micro- and Macro-Level Theories, Processes, and Outcomes. J. Manag. 2014, 40 (5), 1202−1229. (29) Cyert, R.; March, J. G. Behavioral Theory of the Firm, 2 ed.; Wiley-Blackwell: Cambridge, MA, 1992. (30) Kaminsky, J.; Javernick-Will, A. Theorizing the Internal Social Sustainability of Sanitation Organizations. J. Constr. Eng. Manag. 2015, 141 (2), 4014071. (31) Hofstede, G. Cultural differences in teaching and learning. Int. J. Intercult. Relat. 1986, 10 (3), 301−320. (32) Bockstedt, J.; Druehl, C.; Mishra, A. Problem-solving effort and success in innovation contests: The role of national wealth and national culture. J. Oper. Manag. 2015, 36, 187−200. (33) Shane, S. Uncertainty Avoidance and the Preference for Innovation Championing Roles. J. Int. Bus. Stud. 1995, 26 (1), 47−68. (34) Tellis, G. J.; Stremersch, S.; Yin, E. The international takeoff of new products: The role of economics, culture, and country innovativeness. Mark. Sci. 2003, 22 (2), 188−208. (35) Yeniyurt, S.; Townsend, J. D. Does culture explain acceptance of new products in a country? - An empirical investigation. Int. Mark. Rev. 2003, 20 (4), 377−396. (36) Baskerville, R. F. Hofstede never studied culture. Account. Organ. Soc. 2003, 28 (1), 1−14. (37) Kirkman, B. L.; Lowe, K. B.; Gibson, C. B. A quarter century of culture’s consequences: A review of empirical research incorporating Hofstede’s cultural values framework. J. Int. Bus. Stud. 2006, 37 (3), 285−320. (38) McSweeney, B. Hofstede’s model of national cultural differences and their consequences: A triumph of faith-a failure of analysis. Hum. Relat. 2002, 55 (1), 89−118. (39) Myers, M. D.; Tan, F. B. Beyond models of national culture in information systems research. Adv. Top. Glob. Inf. Manag. 2003, 2, 14− 29. (40) Taras, V.; Kirkman, B. L.; Steel, P. Examining the impact of Culture’s consequences: A three-decade, multilevel, meta-analytic review of Hofstede’s cultural value dimensions. J. Appl. Psychol. 2010, 95 (3), 405. (41) World Bank. World Development Indicators, World DataBank. http://databank.worldbank.org/data/reports.aspx?source=worlddevelopment-indicators (accessed Aug 24, 2015). (42) Rousseeuw, P. J.; Van Zomeren, B. C. Unmasking multivariate outliers and leverage points. J. Am. Stat. Assoc. 1990, 85 (411), 633− 639. 7731
DOI: 10.1021/acs.est.6b01107 Environ. Sci. Technol. 2016, 50, 7723−7731
