Water Resources Research

Household water saving: Evidence from Spain

Authors


Corresponding author: G. Larramona, Department of Economic Analysis, Universidad de Zaragoza, Gran Vía, 2 50005 Zaragoza, Spain. (gemmalar@unizar.es)

Abstract

[1] This article focuses on household water use in Spain by analyzing the influence of a detailed set of factors. We find that, although the presence of both water-saving equipment and water-conservation habits leads to water savings, the factors that influence each are not the same. In particular, our results show that those individuals most committed to the adoption of water-saving equipment and, at the same time, less committed to water-conservation habits tend to have higher incomes.

1. Introduction

[2] Although water supply projections predict lower and more unstable precipitation and lower river runoffs due to climate change, demand for water will continue to grow. A clear example of this divergence between supply and demand is Spain, a country where many regions suffered from water shortages, while the level of household water consumption was the highest among European Union (EU) countries in 2000 (see Figure 1).

Figure 1.

Consumption of water in Europe, 2000 (based on data from European Water Association (EWA), Yearbook 2002).

[3] The general position, based on the accumulation of larger water volumes in surplus areas and their transfer to deficit areas, seems to be outdated, and a posture in favor of protection of water ecosystems in the context of sustainable development is becoming prevalent in Europe [European Commission (EC), 2007]. However, a consensus is far from being reached in Spain. The scarcity of water along the Mediterranean coast, together with recent reforms of the Autonomous Statutes (implying an increase in regional government powers over water affairs), explain the persistence of voices in favor of water transfer projects, even if most cost-benefit analyses advise against such schemes [see, for example,Albiac et al., 2003]. The Spanish regions are developing their own legislative initiatives on water policy, and, hence, regional disputes will become more frequent and bitter (see, for instance, Rosenthal [2008].”

[4] One of the proposed measures currently receiving a significant measure of support from a variety of social groups is the efficient use of water. In this article, we focus on the water behavior of Spanish households, using the Survey on Households and the Environment (SHE) 2008, which provides information about the habits, consumption patterns, and attitudes of Spanish households with regard to the environment and, in particular, to water conservation. Our aim is to identify the factors that determine the level of household water-conservation behavior, which in turn will allow us to establish useful policy implications.

[5] Earlier studies have focused on identifying the socioeconomic, demographic, and attitudinal factors that influence household water savings, since defining the characteristics of those individuals who save water helps policy makers in the implementation of consciousness-raising campaigns encouraging conservation attitudes within society. A useful identification of different types of individuals, in terms of water-saving habits, is made byGilg and Barr [2006], using a sample of 1600 households in Devon, United Kingdom. This study also provides evidence that the implementation of policies with other aims (energy conservation, green consumption, recycling) may also affect water-saving behavior. More recently,Millock and Nauges [2010], using survey data from 10 countries of the Organisation for Economic Co-operation and Development (OECD), study the factors affecting the household adoption of water-saving equipment. They find that households subject to a volumetric fee have a much greater probability of investing in water-saving equipment than households that pay a flat fee.Mondéjar-Jimenez et al. [2011]study the water-saving behavior of Spanish households, including in their analysis water-saving habits and equipment, and they, unexpectedly, find that higher education or higher income have a negative effect on water-saving patterns of behavior; this is counter to findings in other countries. Our intuition leads us to consider that this empirical result may be because the adoption of water-saving equipment is connected to factors that are not necessarily the same as those that determine water-saving habits.

[6] In this regard, our article makes two main contributions. First, a theoretical model that considers the optimal choice of a representative consumer about the purchase of water-saving equipment and the adoption of water-saving habits, in which we assume that the adoption of such habits does not require expenditures but does imply an opportunity cost in terms of time. On the contrary, the purchase of water-saving equipment uniquely involves a payout. Our theoretical analysis identifies different patterns among these variables; in particular, those individuals whose time is more valuable, that is to say, those with higher incomes will choose to acquire water-saving technology rather than adopt water-saving habits, because the cost in relative terms of the former is lower.

[7] The second contribution consists of an empirical analysis carried out in three steps. Following Mondéjar-Jimenez et al., we begin our empirical study by analyzing different water-saving strategies jointly. We choose an alternative method to that of these authors by employing the water-saving indicator (WSI), designed by the Spanish Statistical Institute (INE), as a dependent variable. This indicator is based on nine variables reflecting the two basic modes of domestic water use: household adoption of water-saving equipment (single or thermostatic taps, water-economy devices such as tap aerators, spraying or reduced-volume mechanisms, and mechanisms to restrict cistern discharge) and water-saving habits (water recycling, filling the kitchen sink before washing up, using full washing and dish-washing machines, semiclosing the stopcock to reduce the flow to taps, turning off the tap while brushing teeth, and lathering and showering rather than taking a bath). The advantage of this index is that it ranks individuals by their water-saving behavior. Departing from Mondéjar-Jimenez et al., we find that the level of income positively affects water-saving behavior, whereas the level of education has a nonlinear effect. These contradictory results cannot be explained by the data, simply because we make use of the same data. This inconsistency is disentangled in the second step of our empirical analysis: we build two indicators that separately reflect the ownership of water-saving equipment and the extent of water-saving habits, called the water-saving equipment indicator (WSEI) and the water-saving habit indicator (WSHI), respectively. These indicators allow us to discover whether variables that determine the household adoption of water-saving equipment are the same as those that influence water-saving habits. We find that the adoption of water-saving equipment obeys factors different from those that determine water-saving behavior. Specifically, we confirm that those most committed to the adoption of water-saving equipment, but less committed to water-saving behavior, tend to have higher incomes. The last step of our empirical study consists in exploring the relationships among different water-saving decisions.

[8] This article has the following structure. In section 2, the model and its main results are presented. Section 3 shows the empirical analysis. Our main conclusions are set out in section 4.

2. The Model

[9] Let us consider an individual or a household deriving utility from goods consumption c and water saving s. The level of water saving can be the result of the installation of water-saving equipmentk that saves inline image cubic meters of water (for instance, mechanisms to restrict cistern discharge or tap aerators) and the allocation of time to good habits for water saving, h which saves inline image cubic meters of water (for instance, time devoted to recycling water or time waiting to fill the kitchen sink before washing up).

[10] The time spent in water-saving behavior is valuable time that individuals could use to generate greater labor income. Total time availableT is for the only two possible activities, work land good water-saving habitsh:

display math

[11] The labor income is the product of the wage rate w and the working time l. Labor income is spent on consumption goods and water-saving equipment. We assume that adopting water-saving habits does entail an opportunity cost in terms of time but does not imply a monetary cost. Let the consumption good be the numeraire good and letψbe the water-saving equipment price in annualized terms. Since the water-saving equipment is a durable good investment, if the equipment is replaced everyq years, ψ would be equal to (r/(1 + r))(p/(1 − (1 + r)q)), with r the individual discount rate and p the purchase price of the equipment [Hausman, 1979]. Therefore, each individual or household faces to the following budget constraint:

display math

[12] The time constraint and budget constraint can be summarized as follows:

display math

in such a way that the individual faces the following optimization problem:

display math

with inline image, inline image, inline image, inline image, inline image, inline image, inline image, inline image, inline image, inline image, and inline image.

[13] This constrained maximization problem can be summarized as expression (2):

display math

[14] The solution meets the following necessary conditions:

display math
display math

[15] From the above expressions, we obtain that, at the optimal allocation, water-saving habits and water-saving equipment have identical marginal benefit/price ratios:

display math

[16] Differentiating expression (5) with respect to h and w, the following expression is obtained:

display math

implying that an increase of the wage rate (the opportunity cost of time) leads to a reduction of time devoted to good water-saving habits. This indicates that those with higher incomes, whose time has a greater opportunity cost, are less committed to spend time in water-saving habits. Furthermore, differentiating expression(5) with respect to k and w, we obtain a negative relationship between both variables:

display math

[17] This expression shows that those with higher incomes are most committed to the adoption of water-saving equipment to save water. These relationships between income and water-saving actions will be empirically analyzed in the next section.

3. Empirical Evidence

3.1. Data Set

[18] Our data are taken from the SHE 2008, which collects information on the practices and behavior of Spanish families in relation to environmental protection. There are two basic units of observation in the survey: the main family dwellings and their residents 16 years old and over. The INE used a stratified sampling and determined that the number of observations required to make inferences at the national level was 14,000 dwellings, in each of which an individual was selected to provide information about household and selectee characteristics. The ideal informant is the reference person of the dwelling, who is the owner of the dwelling or head of household. Where this was not possible, the selected person would be another adult member of the household. A complete methodological note is available in Mas and Iztueta-Azkue [2009]. The microfile consists of 26,689 observations. Responses without answers to the income and environmental concern questions reduce the number of observations to 19,646, ensuring a representative sample.

[19] In the first part of our analysis, we use WSI as a dependent variable. This indicator is based on nine variables reflecting different water-saving actions: having single or thermostatic taps in the house, having water-economy devices such as tap aerators, spraying, or reduced-volume mechanisms in the house, having mechanisms to restrict cistern discharge, water recycling, filling the kitchen sink before washing up, using full washing and dish-washing machines, semiclosing the stopcock to reduce the flow to taps, turning the tap off while brushing teeth, and lathering and showering rather than taking a bath. Except for the latter two, all these variables refer to the dwelling and the behavior of the members of the dwelling, respectively. Turning the tap off while brushing teeth, and lathering and showering rather than taking a bath refer exclusively to the selectee. Although the survey questionnaire explicitly points out that respondents should answer “yes” or “no” according to the behavior of the members of the dwelling, extrapolating from the characteristics of one respondent to the entire household's water-saving behavior could be problematic. Moreover, although the respondent provides information on whether or not the house is equipped with devices/mechanisms for saving water, he/she may not be the one responsible for installing this equipment, or the decision may have been taken some years ago in such a way that the socioeconomic and demographic characteristics of the dwelling at the moment of answering the questionnaire may be different from those at the moment of purchasing the water-saving equipment. However, it is impossible to glean this information from the survey.

[20] Table 1 shows the numerical values given for each response. The variables are recoded taking values of −1, 0, or 1, following INE guidelines, and are then weighed together in the index WSI. Table 1 shows the relative importance of each variable considered, which comes from the mean of the different proposals of the various research groups who worked on the project. This indicator has been developed by a group of experts selected by INE and several regional statistics institutes (Institute of Statistics and Cartography of Andalusia, the Statistical Institute of Catalonia, and the Statistical Institute of Galicia). The robustness of the indicator was tested by recalculating the indicator with proposals from each group and finding no significant variations.

Table 1. Water-Saving Use Indicator: Variables and Weightsa
VariablesWeight
  • a

    The variables SHABIT1 and SHABIT2 are not answered by 8% of the interviewees.

DISP1: single or thermostatic tap (yes = 1, no = 0)0.0825
DISP2: water-economization devices (yes = 1, no = 0)0.1252
DISP3: mechanisms to restrict cistern discharge (yes = 1, no = 0)0.1252
HABITO1: water recycling (yes = 1, no = −1)0.1175
HABITO4: filling the kitchen sink before washing up (yes = 1, no = −1)0.0821
HABITO5: using full washing and dish-washing machines (yes = 1, no = −1)0.1045
HABITO6: semiclosing the stopcock to reduce the flow to taps (yes = 1, no = −1)0.1001
SHABIT1: turning the tap off while brushing teeth or lathering (yes = 1, no = −1)0.1390
SHABIT2: showering instead of taking a bath (yes = 1, no = −1)0.1239

[21] Figure 2presents the percentage of positive and negative responses given by the interviewees to every variable considered in the indicator. In five out of the nine components considered, the majority of the responses are negative. The positive responses have to do with having single or thermostatic taps, using full washing and dish-washing machines, turning the tap off while brushing teeth, or lathering and showering rather than taking a bath. The factors relative to water recycling or water-economy devices produce, in the main, negative responses.

Figure 2.

Descriptive statistics of the indicator components.

[22] To obtain positive values for the indicator, we transform the range from 0 to 10. We give a code number, 0, 5, or 10, to the values for each response −1, 0, or 1, respectively. Therefore, for the case of HABITO1, the response yes would have value 10 and the response no would have value 0.

[23] In this way, values close to zero WSI indicate low levels of water conservation, while values close to 10 indicate the opposite. The mean of the indicator is 6.3 for the 19,646 individuals interviewed, with a standard deviation of 1.3.

[24] In the second part of the empirical analysis, the WSEI and the WSHI are created to separately reflect the use of water-saving equipment and the extent of water-saving habits. In order to decompose the WSI to two indices and maintain the weights provided by INE to elaborate the WSI, a rule of three has been used. Both indices have been normalized to 10. For instance, fromTable 1 it can be seen that HABITO1 has a weight of 0.1175 in the WSI. The HABITO1 is also one of the six components in WSHI whose sum of the weights provided by INE is 0.6771; therefore, we use the rule of three to transform its weight to 0.1735 in the WSHI and obtain one as a maximum value of the index.

[25] The different water-saving actions used to build the previous indices are also explored. Looking at the conditional probabilities (Table 2), it can be seen that some water-saving actions are linked to each other. Mainly, the presence of one kind of water-saving equipment in the dwelling increases the probability of having another. In particular, the presence of water-saving devices significantly increases the probability of having single or thermostatic taps. Interestingly, the same occurs with respect to water-saving habits; the presence of one habit increases the probability of having another. However, we observe that, in those dwellings with water-saving equipment, the probability of filling the kitchen sink before washing up decreases considerably.

Table 2. Conditional Probabilities of the Water-Saving Actions
Ip(i)p(i|DISP1=yes)p(i|DISP2=yes)p(i|DISP3=yes)p(i|HABITO1=yes)p(i|HABITO4=yes)p(i|HABITO5=yes)p(i|HABITO6=yes)p(i|SHABIT1=yes)p(i|SHABIT2=yes)
  • a

    DISP, HABITO, and SHABIT refer to variables used by the Spanish Statistics Institute (INE). They are used here to facilitate locating the variables in the INE website.

DISP1ba0.669710.84540.84100.68720.65060.70870.67000.67670.6739
DISP2b0.14450.182410.27350.18290.14160.16290.17450.14830.1463
DISP3b0.30900.38800.584710.36680.30790.34650.33920.31700.3126
HABITO1b0.20290.20820.25680.240910.24950.21420.27390.21370.2041
HABITO4b0.38400.37300.37620.38270.472210.40830.44880.39110.3843
HABITO5b0.81190.85910.91510.91050.85730.863210.84040.81880.8156
HABITO6b0.28760.28780.34730.31580.38840.33620.297710.29670.2890
SHABIT1b0.90250.91200.92640.92590.95080.91910.91030.930810.9108
SHABIT2b0.96440.97040.97600.97560.97010.96530.96890.96890.97331
Number of observations19,64613,157283960703986754415,950565117,73118,947

[26] Apart from the explanatory variables considered in the model to analyze water-saving behavior, prior research also considers the influence of several independent variables. In the following, we present all the independent variables used in the empirical estimation.

3.1.1. Socioeconomic and Demographic Variables

[27] At the microlevel, we consider gender, age, and years of education of the selectee, household income, and household type proxied by living with dependent children and number of people living in the house. Some of these variables do have an ambiguous influence in prior studies, and this is certainly the case with gender. Gilg and Barr [2006] establish that those less committed to water savings in the home tend to be male, whereas Millock and Nauges [2010] find that gender is not significant.

[28] Gilg and Barr [2006]also find evidence of a positive relationship between age and water saving, though several studies have stressed that age is negatively correlated with a willingness to contribute to additional environmental protection, since older people will not live long enough to enjoy the long-term benefits of preserving resources [Carlsson and Johansson-Stenman, 2000].

[29] The type of household may have an influence on water conservation. Hines et al. [1987], and Gilg and Barr [2006], report that water conservation is linked to smaller family size. This is contrary to the findings of Millock and Nauges [2010]. The arrival of children makes the future “a far more tangible concept” and causes individuals to reconsider present behavior in light of future consequences [Dresner et al., 2007], resulting in households with children being more concerned about the environment and the future and, hence, being more likely to adopt water-saving behaviors. However,Torgler et al. [2008] were not able to observe that having children is positively correlated with a stronger preference toward the environment.

[30] Different results are found when we focus on income. Berk et al. [1993] and Gilg and Barr [2006] find a significant and positive effect of household income on water conservation. Domene and Sauri [2006] find no significant relationship, whereas Mondéjar-Jimenez et al. [2011] show a negative effect. Renwick and Archibald [1998] and Millock and Nauges [2010]observe a positive influence of income on the probability of installing indoor water-saving equipment, and a negative effect on the probability of buying a water tank.

[31] With respect to education, Gilg and Barr [2006] find a positive influence of education on household water savings, Lam [2006]finds that higher education does not necessarily affect adoption of dual-flush controllers, andMillock and Nauges [2010] do not include this variable, considering that education level is correlated with income. Mondéjar-Jimenez et al. [2011] detect a negative effect of education on water saving in Spanish households.

3.1.2. Preference Variables Toward the Environment

[32] Following Millock and Nauges [2010], who claim that a strong commitment to environmental values influences household adoption of water-saving equipment, we consider that whatever variable affects attitude toward the environment is likely to have an influence on water-saving behavior. The survey provides information about the environmental awareness of the respondent. First, the variables detection of environmental problems, worry about environment, and being acquainted with an environmental campaign are dummies, introduced to capture individual environmental awareness in general terms. Collaborating with an environmental protection organization, participating as an environmental volunteer, signing against situations that damage the environment, demonstrating against a situation that damages the environment, and making a formal complaint about an environmental problem may be proxies for the extent of involvement in environmental activities.Torgler and García-Valiñas [2007] use the variable membership in voluntary organizations as a social capital proxy, finding that individuals who participate actively in environmental organizations have stronger preferences for environmental protection. These authors also make utility a variable of trust (whether most people can be trusted or not), but this information is not available to us. The survey also provides information about the importance given to a particular element when choosing a product (price, brand, efficiency, green quality, and localization of production), which may indirectly reflect environmental preferences. Prior studies identify purchasing locally produced foods, and green consumption, as sustainable consumption [Gilg et al., 2005], and Millock and Nauges [2010]show that the habit of purchasing green products positively increases the probability of adopting indoor water-saving equipment. Finally, the following variables, being in favor of specific environmental measures as obliging separation of domestic rubbish under threat of fine, regulating or restricting abusive consumption of water in households, establishing an environmental tax for polluting fuels, establishing restrictions on the use of private transport, establishing an ecological tax on tourism, installing a renewable energy farm in the municipality, paying more for use of alternative energy, and decreasing noise on the main streets, allow us to measure individual environmental sensitivity toward specific subjects like alternative energy use and recycling.

[33] The list of variables considered in the following econometric analysis and their descriptive statistics are given in Appendix A.

3.2. Empirical Results

3.2.1. WSI Indicator as a Dependent Variable

[34] Table 3 provides the estimation results. Except for the variable income, the interpretation of the coefficients is that an increase in one unit in the exogenous variable increases the index in the relevant units. For instance, as Table 3 shows, an increase of one unit in the number of people living in the house increases the WSI by 0.058 units. If the variable is binary, the interpretation is, for instance, that being female increases the WSI by 0.0758 units. Given that the variable income is in logarithms, the interpretation is that a 1% increase in income increases the WSI by 0.000537 units (see Table 3). We have maintained certain statistically insignificant variables to reduce the likelihood of inconsistent parameter estimation due to omitted-variable bias. To ensure that the conditional mean is correctly specified, we apply a Ramsey RESET test, concluding that the model has no omitted variables. The null hypothesis that the model has no relevant omitted variables cannot be rejected at the 5% level of significance. Applying the Breusch-Pagan test for heteroskedasticity in previous ordinary least-square regressions, we detect its presence, leading us to estimate the parameters of the model by using the two-step feasible generalized least-square estimation method, obtaining potential gains in efficiency.

Table 3. Feasible Generalized Least Squares Estimation Resultsa
Socioeconomic and Demographic FactorsWSIRobust Standard Errorp > |t|
  • a

    Dependent variable: WSI.

Women0.07580.01770.000
Age0.01920.00280.000
Age2−0.00020.00000.000
Years of qualification0.02420.00860.005
Years of qualification2−0.00080.00040.055
Log income0.05370.02140.012
With dependent children−0.04830.02530.056
Number of people living in the house0.05800.01000.000
Preferences toward environment   
 Concern about environment0.19560.02360.000
 Knowledge of environmental campaign0.19550.01950.000
 Detection of an environmental problem0.04940.02130.020
 Organizational collaboration0.12900.04890.008
 Participating as an environmental volunteer0.09130.05580.102
 Signing against environmental damages0.14990.02910.000
 Demonstrating against a situation harmful to the environment−0.00760.04310.860
 Complaint against environmental damages0.07790.05080.125
 Price0.12030.03460.001
 Brand−0.09410.01790.000
 Efficiency0.23520.02360.000
 Green0.15570.01970.000
 Local0.07580.01850.000
 Penalty for not recycling0.05940.01930.002
 Control for abuse of water use0.17470.02520.000
 Environmental tax for polluting fuels0.05070.02090.015
 Restrictions for private transport0.08620.01980.000
 Ecological tax for tourism0.06490.02050.002
 Renewable energy farm0.28880.02170.000
 Paying more for alternative energy0.00750.02170.729
 Decreasing noise on the main streets0.12560.02770.000
 Constant3.92160.16750.000
 No observations19612  
 R20.1091  
 Ramsey test (Ho: model has no ommited variables)0.054  

[35] Most regressors are highly statistically significant, with the expected coefficient signs. However, we find effects of income and education on water-saving behavior different from those found byMondéjar-Jimenez et al. [2011]using the same data. In our analysis, education has a positive effect, but its influence almost disappears at higher levels. Income emerges as a variable with positive effect. The explanation of this divergence may be given by the theoretical model, which indicates that income affects positively (negatively) the adoption of water-saving equipment (the implementation of water-saving habits). We check this in the following subsections.

3.2.2. WSEI and WSHI as Dependent Variables

[36] Equations (6) and (7)suggest that the amount of water-saving equipment in a dwelling depends positively on the level of income, whereas the implementation of water-saving habits may depend negatively on the level of income.Table 4 provides the estimation results using WSEI and WSHI as dependent variables. First, we can complete the portrait highlighted by Gilg and Barr [2006]of those most committed to water-saving behavior. We confirm that committed individuals tend to be female and older, although we note that the effects of age are nonlinear. An inverse U-shaped relationship is detected between age and water-saving actions, which may indicate that, at much greater ages, concern about the future diminishes. According to the findings ofMillock and Nauges [2010], we confirm that the greater the number people living in the same house, the higher the probability of the presence of water-saving equipment, and the greater the number of individuals in the household, the more likely will be the adoption of water-saving habits.

Table 4. Feasible Generalized Least Squares Estimation Resultsa
Socioeconomic and Demographic FactorsWSEIRobust Standard Errorp > |t|WSHIRobust Standard Errorp > |t|
  • a

    Dependent variables: WSEI and WSHI.

Women0.08470.03800.0310.09000.02340.000
Age0.03900.00360.0000.01790.00370.000
Age2−0.00040.00000.000−0.00010.00000.000
Years of qualification0.11820.02150.0000.00290.01140.798
Years of qualification2−0.00300.00100.001−0.00030.00060.602
Log income0.66300.19380.000−0.09500.02830.001
With dependent children0.13090.06650.024−0.11210.03330.001
Number of people living in the house0.03100.03310.1690.08160.01330.000
Preferences toward environment      
 Concern about environment0.12300.06870.0130.26860.03160.000
 Knowledge of environmental campaign0.31730.15180.0000.21560.02610.000
 Detection of an environmental problem0.25040.08520.0000.01160.02800.678
 Organizational collaboration0.58710.12540.0000.04690.06370.462
 Participating as an environmental volunteer0.24420.14870.0730.07740.07280.287
 Signing against environmental damages0.33180.09300.0000.14840.03810.000
 Demonstrating against a situation harmful to the environment−0.12860.10940.2250.02350.05630.676
 Complaint against environmental damages0.12670.12120.3350.08440.06620.202
 Price−0.06750.06790.3700.18560.04730.000
 Brand−0.20710.10810.000−0.07790.02360.001
 Efficiency0.47580.11550.0000.22850.03130.000
 Green0.03960.05610.3730.22820.02600.000
 Local0.25520.08680.0000.04510.02440.065
 Penalty for not recycling0.06270.08900.1460.07310.02550.004
 Control for abuse of water use0.06230.08110.2450.24310.03340.000
 Environmental tax for polluting fuels−0.03940.08610.3960.08990.02770.001
 Restrictions for private transport0.01900.06700.6690.12750.02610.000
 Ecological tax for tourism0.01440.09030.7570.08910.02700.001
 Renewable energy farm0.51490.09910.0000.29450.02870.000
 Paying more for alternative energy−0.10810.09620.0310.03030.02860.291
 Decreasing noise on the main streets0.27990.08760.0000.11780.03770.002
 Constant−4.61171.37140.0004.65790.22200.000
 No observations19,612  19,612  
 R20.1299  0.0728  
 Ramsey test (Ho: model has no ommited variables)0.047  0.076  

[37] The empirical results confirm the theoretical hypothesis. Higher incomes have a positive effect on the probability of the presence of water-saving equipment, while water-saving habits decrease. Education maintains its influence on the use of water-saving equipment, but this influence disappears with respect to water-saving habits. Counterintuitively, skilled individuals seem to have the same water-saving habits as the unskilled.

[38] We also find that the type of household affects water conservation. Households with dependent children have fewer water-saving habits, but more water-saving equipment. One explanation for this could be that child care requires time, which may increase the competition for the resource time between child care and having water-saving habits. Another explanation is that younger children are not concerned about water saving; for instance, they would generally rather have a bath than a shower. The opposite effect on equipment could be due to the fact that households with children tend to live in new or refurbished houses with modern water-saving equipment. Furthermore, Spanish individual households tend to be less efficient; that is to say, they require more resources per capita than larger households. This is in line with a report on EU water-saving potential by the Ecological Institute for International and European Environmental Policy in 2007, which estimated that a two-person household uses around 300 L of water per day, whereas two single households use 210 L each.

[39] Not surprisingly, most variables reflecting individual environmental preferences have a positive impact on WSEI and WSHI. It is worth noting that, if the main concern in choosing a product is the price, the green characteristics of the product, the source of the production, or the efficiency of the product, the WSHI increases. The latter concern also has an important positive effect on WSEI. Against this, if the main concern is brand, the WSEI and the WSHI decrease.

[40] At this stage, it could be interesting to analyze all the water-saving actions individually, in order to discover whether more differences emerge.

3.2.3. WSI Components as Dependent Variables

[41] Table 5provides the estimation results of the WSI components as dependent variables, using multivariate analysis. The purpose of multivariate analysis is to treat multivariate data as a whole, summarizing the data and revealing structure. Due to the fact that the different water-saving actions are not independent actions, asTable 2 shows, the multivariate probit model seems to be more appropriate. The previously independent variables are also used in the multivariate probit model. It is necessary to clarify that, due to the lack of a clear interpretation of the squared variables in binary models, the age and qualification variables have been replaced by a set of dummy variables to capture nonlinearity. In the case of age, four dummy variables have been introduced reflecting the following age range: 16–29 years, 30–44 years, 45–64 years, and more than 64 years, with the range 16–29 being the control variable. In the case of education, four dummies have been considered: illiterate individuals, and those with primary, secondary, or tertiary education, using illiterate as the control variable. Table 6 shows the statistically significant and positive estimated correlation coefficients (ρij), revealing that the unobservable factors increasing the probability of adopting one type of water-saving equipment also affect the probability of having another type. Furthermore, the likelihood test for independence between the disturbances is strongly rejected, implying correlated binary responses between different water-saving actions.

Table 5. Multivariate Analysisa
Socioeconomic and Demographic FactorsDISP1p>|t|DISP2p>|t|DISP3p>|t|HABITO1p>|t|HABITO4p>|t|HABITO5p>|t|HABITO6p>|t|SHABIT1p>|t|SHABIT2p>|t|
  • a

    Dependent variables: components of the WSI.

Women0.01560.418−0.00280.9040.03920.0430.08320.0000.04150.0260.05740.008−0.03320.0860.12670.000−0.09200.009
Age0.18380.0000.06780.0680.14080.0000.05570.121−0.08190.0090.08480.0230.02490.4460.21490.0000.10910.065
Age 30–440.11120.0010.05360.1640.08580.0090.11200.0020.02330.4690.08820.0210.07550.0240.16820.0000.05300.380
Age 45–640.04930.1900.00260.955−0.02550.5060.09770.0190.07570.0390.09660.0230.11710.0020.17410.000−0.03750.577
Age > 640.12700.0260.07800.357−0.01640.803−0.13480.033−0.06210.2710.23460.000−0.16540.005−0.17430.0340.16670.057
Primary school0.28890.0000.26200.0020.16330.015−0.06770.303−0.19400.0010.25370.000−0.09130.137−0.18090.0350.10350.273
Secondary school0.326900.27830.0020.20660.004−0.12120.086−0.300700.30590−0.11710.074−0.14170.1220.11990.255
Tertiary school0.36430.0000.18370.0000.19620.000−0.23070.000−0.06150.0060.23630.000−0.04210.070−0.07300.0170.12020.005
With dependent children0.01880.5040.05500.0910.03170.250−0.06500.031−0.08660.0010.07460.022−0.08080.004−0.04340.232−0.02430.640
Number of people living in the house0.01560.153−0.00940.4790.02380.0300.06220.0000.03490.0010.06540.0000.03580.001−0.01810.197−0.00310.875
Preferences toward environment                  
 Concern about environment0.08830.0000.01600.6070.08230.0010.08800.002−0.03140.1920.07120.0080.13380.0000.32160.0000.13670.001
 Knowledge of environmental campaign0.10870.0000.11940.0000.12470.0000.08050.0000.09480.0000.18250.0000.05240.0140.09440.0010.09800.009
 Detection of an environmental problem0.05420.0230.04130.1210.12200.0000.11010.000−0.04070.072−0.09440.0000.05000.031−0.03320.295−0.04190.337
 Organizational collaboration0.18510.0010.22410.0000.11670.0240.08830.112−0.01280.8090.03580.5980.01570.7660.02730.726−0.15020.134
 Participating as an environmental volunteer−0.04130.5190.04140.5250.16370.0050.02190.731−0.02410.6870.00040.9960.10950.0670.06990.430−0.07820.493
 Signing against environmental damages−0.01100.7410.12420.0000.14260.0000.09440.005−0.06410.0410.12500.0020.11060.0000.09360.0430.01010.874
 Demonstrating against a situation harmful to the environment−0.11080.023−0.05400.297−0.01160.800−0.01180.8120.07240.118−0.09860.0880.02780.5500.00290.9660.13950.170
 Complaint against environmental damages−0.11580.0450.10780.0700.09220.0870.14480.011−0.14910.0070.08990.2000.11290.038−0.05260.5030.11820.327
 Price−0.01880.618−0.00360.937−0.03450.359−0.05480.1790.17780.0000.12730.0020.04250.2640.10790.0220.09720.124
 Brand0.02020.300−0.11960.000−0.10840.000−0.08640.000−0.03030.106−0.02590.2410.03200.101−0.07530.004−0.08840.013
 Efficiency0.13600.0000.25600.0000.19210.0000.07160.0120.07110.0040.28360.0000.00510.8440.08810.0060.01420.741
 Green−0.01710.4280.00830.7410.02960.1660.12260.0000.09490.0000.04640.0580.14980.0000.05990.0360.13170.001
 Local0.01450.4720.15990.0000.10240.0000.06990.0010.03300.0890.12710.000−0.08100.000−0.05480.0400.04490.218
 Penalty for not recycling0.00970.6480.04300.0850.01440.4980.13030.0000.00580.778−0.04330.0710.03310.1190.03460.215−0.07020.070
 Control for abuse of water use0.06380.0150.00510.8730.02130.4290.17160.0000.02030.4290.15840.000−0.01850.4880.21250.0000.13360.003
 Environmental tax for polluting fuels−0.03580.1170.00050.985−0.00180.937−0.00720.7740.04430.0440.16980.0000.01560.497−0.03120.2960.00930.822
 Restrictions for private transport0.01100.614−0.03270.1970.02910.1770.12220.0000.04300.039−0.02410.3290.07320.0010.04450.1250.04290.284
 Ecological tax for tourism0.04900.0310.06770.009−0.06880.0020.02890.2290.04500.0370.02050.4320.07370.0010.00960.752−0.02570.539
 Renewable energy farm0.13320.0000.20430.0000.22740.0000.09410.0000.10970.0000.24090.0000.07770.0010.12960.0000.08390.035
 Paying more for alternative energy0.03260.180−0.05910.033−0.05360.024−0.00060.9820.05440.018−0.00130.963−0.00420.8590.0380.2500.01490.743
 Decreasing noise on the main streets0.12320.0000.06660.0720.13720.000−0.02380.467−0.02050.4690.15340.0000.01720.5650.0750.0380.15610.001
 Constant−3.04820.000−3.2290.000−2.8130.0000.10890.571−0.23510.167−2.39460.000−0.6540.0001.07450.0000.29250.353
Table 6. Correlation Coefficients Between Different Saving Actions, ρij, Where ija
 ρijp > |t| ρijp > |t|
  • a

    Likelihood test of all ρij = 0: χ2836 = 2886.52 Prob > χ2=0.000.

  • a

    We denote DISP1 as 1, DISP2 as 2, DISP3 as 3, HABITO1 as 4, HABITO4 as 5, HABITO5 as 6, HABITO6 as 7, SHABIT1 as 8, and SHABIT2 as 9.

ρ210.24210.000ρ730.04530.000
ρ310.31260.000ρ830.07900.000
ρ410.01680.189ρ930.07440.001
ρ51−0.03490.002ρ540.12280.000
ρ610.19150.000ρ640.08200.000
ρ71−0.02120.072ρ740.15530.000
ρ810.04530.003ρ840.14090.000
ρ910.06630.001ρ940.01260.571
ρ320.32520.000ρ650.17060.000
ρ420.08630.000ρ750.12360.000
ρ520.00540.671ρ850.07710.000
ρ620.15160.000ρ950.00680.738
ρ720.06210.000ρ760.03940.002
ρ820.04880.006ρ860.03920.016
ρ920.03690.128ρ960.05560.011
ρ430.07780.000ρ870.10080.000
ρ530.01010.370ρ970.01240.557
ρ630.22300.000ρ980.31850.000

[42] Higher income and education have a positive effect on the probability of adoption of single or thermostatic taps, water-economy devices, and mechanisms to restrict cistern discharge, whereas those same factors decrease the probability of implementing water-saving habits, except for the habits using full washing and dish-washing machines and having a shower instead a bath, in which the effect of income is positive. Notice that both actions do not imply a time cost, as in the case of water recycling; on the contrary, they lead to time savings.

[43] Households with dependent children have a lower probability of following water-saving habits, except for the habit using full washing and dish-washing machines, since a greater number of household members make it more likely that washing and dish-washing machines will be full. These households have a high probability of investing in water-economy devices.

[44] More interestingly, the habit of purchasing efficient products in general increases the probability of adopting water-saving equipment, whereas it does not affect the probability of some water-saving habits among Spanish households (semiclosing the stopcock to reduce flow to the taps and showering rather than taking a bath). The probability of adopting these habits increases if households tend to purchase green products.

[45] Valuable information comes from the marginal effects in determining the most influential factors for every component of the index, although this information cannot be easily extracted from the multivariate model. It would be necessary to compute the conditional probability that a component were 1 while the rest of the components of the index were 0, but this cannot currently be computed using the programs designed by Capellari and Jenkins, 2003, for Stata (see http://www.stata.com/statalist/archive/2003-11/msg00634.html). An alternative is to estimate the components of the WSI separately with probit models. Table 7 shows these estimations and the results are almost equal in values and significance to Table 5. These estimations do not suffer heteroskedasticity, since the likelihood-ratio test of heteroskedasticity, which tests the full model with heteroskedasticity against the full model without, is not significant at the 1% level.Table 8 provides the marginal effects.

Table 7. Probit Estimation Resultsa
Socioeconomic and Demographic FactorsDISP1p>|t|DISP2p>|t|DISP3p>|t|HABITO1p>|t|HABITO4p>|t|HABITO5p>|t|HABITO6p>|t|SHABIT1p>|t|SHABIT2p>|t|
  • a

    Dependent variables: components of WSI indicator.

Women0.01550.423−0.00460.8430.03680.0590.08320.0000.04130.0260.05720.009−0.03430.0760.12120.000−0.09630.006
Age 30–440.18100.0000.06910.0650.14100.0000.05890.102−0.08200.0090.08380.0260.02370.4700.21280.0000.09740.102
Age 45–640.11250.0010.05290.1710.08150.0140.11300.0020.02340.4670.08780.0220.07370.0280.16210.0000.04330.477
Age > 640.05100.1760.00200.966−0.03370.3830.09710.0200.07440.0420.09820.0210.11520.0030.17370.000−0.04090.546
Primary school0.12340.0300.08710.306−0.01950.767−0.13190.037−0.06340.2620.23960.000−0.16350.006−0.16200.0480.18030.039
Secondary school0.28980.0000.26260.0020.15370.023−0.06690.310−0.19570.0010.26490.000−0.08950.146−0.16590.0520.11980.204
Tertiary school0.32470.0000.28630.0020.20180.005−0.11980.090−0.30300.0000.31120.000−0.11430.082−0.12670.1660.13120.212
Log income0.36490.0000.18050.0000.19360.000−0.22920.000−0.06120.0060.23660.000−0.04260.067−0.07260.0170.11890.006
With dependent children0.01890.5030.05640.0830.03110.260−0.06710.026−0.08750.0010.07350.025−0.08130.004−0.04290.239−0.02360.652
Number of people living in the house0.01570.151−0.01190.3640.02120.0530.06100.0000.03480.0010.06460.0000.03530.001−0.01750.217−0.00170.933
Preferences toward environment                  
 Concern about environment0.08850.0000.01160.7100.08210.0020.08700.002−0.03190.1850.07310.0070.13240.0000.31760.0000.13160.002
 Knowledge of environmental campaign0.10870.0000.11370.0000.12270.0000.07690.0010.09470.0000.18400.0000.05230.0140.09690.0000.10410.006
 Detection of an environmental problem0.05360.0250.03940.1420.12010.0000.11060.000−0.04310.057−0.08830.0010.04900.035−0.03120.328−0.04030.359
 Organizational collaboration0.19140.0010.22860.0000.11940.0220.08950.108−0.01260.8110.02740.6870.01240.8150.02280.770−0.15690.118
 Participating as an environmental volunteer−0.04490.4840.03890.5530.15790.0080.01920.763−0.02340.696−0.00160.9830.11040.0650.06540.461−0.07690.503
 Signing against environmental damages−0.01190.7200.12150.0010.14100.0000.09340.005−0.06510.0380.13320.0010.11070.0000.10020.0310.02420.706
 Demonstrating against a situation harmful to the environment−0.11590.017−0.04510.385−0.00720.877−0.01000.8390.07350.113−0.10990.0590.02910.5310.00330.9620.14030.171
 Complaint against environmental damages−0.11580.0440.10920.0670.08480.1170.14250.013−0.14840.0080.09960.1620.11110.042−0.06300.4210.09560.425
 Price−0.02150.5690.00570.899−0.02930.441−0.05320.1920.18020.0000.12770.0020.04540.2330.10750.0230.09480.136
 Brand0.02110.279−0.12670.000−0.11110.000−0.08460.000−0.03050.104−0.02350.2900.03260.095−0.07780.003−0.08480.017
 Efficiency0.13490.0000.26090.0000.19430.0000.07190.0110.07150.0040.27930.0000.00380.8820.08240.0100.00340.937
 Green−0.01500.4880.00740.7690.02910.1760.12430.0000.09580.0000.04570.0630.15220.0000.05700.0470.12750.001
 Local0.01140.5720.16250.0000.10310.0000.07060.0010.03150.1040.12410.000−0.08080.000−0.05490.0400.04330.236
 Penalty for not recycling0.01140.5910.03600.1510.01450.4970.13140.0000.00670.741−0.03980.1000.03430.1060.03690.188−0.06540.093
 Control for abuse of water use0.06590.012−0.00110.9720.01460.5890.17000.0000.01800.4830.15750.000−0.02120.4270.21300.0000.13560.003
 Environmental tax for polluting fuels−0.03510.123−0.00100.972−0.00230.920−0.00730.7710.04490.0420.17230.0000.01620.482−0.03360.2610.00750.856
 Restrictions for private transport0.01220.576−0.03630.1540.02870.1870.12580.0000.04250.042−0.02110.3970.07180.0010.04360.1340.04550.26
 Ecological tax for tourism0.04700.0380.07070.007−0.06990.0020.02990.2140.04480.0380.02030.4380.07550.0010.01120.714−0.02620.533
 Renewable energy farm0.13310.0000.19670.0000.22320.0000.08840.0000.10680.0000.24790.0000.07550.0010.13620.0000.09400.018
 Paying more for alternative energy0.03400.162−0.05660.042−0.05230.0290.00070.9770.05590.0150.00130.965−0.00210.9280.03790.2520.01050.819
 Decreasing noise on the main streets0.12070.0000.06550.0780.13490.000−0.02350.474−0.02050.4690.15340.0000.01790.5490.07960.0270.16720
 Constant−3.04820.000−3.19060.000−2.77000.0000.10120.598−0.23200.173−2.41000.000−0.64900.0001.06140.0000.28390.371
 No observations19,646 19,646 19,646 19,646 19,646 19,646 19,646 19,646 19,646 
 R20.06 0.05 0.06 0.03 0.02 0.08 0.02 0.04 0.03 
 Correctly predicted69.19% 85.55% 69.62% 79.72% 61.74% 81.66% 71.23% 90.25% 96.44% 
 Pearson test (Ho: model has no ommited variables)0.134 0.135 0.103 0.326 0.331 0.297 0.293 0.257 0.276 
Table 8. Marginal Effects of Probit Models
Socioeconomic and Demographic FactorsDISP1DISP2DISP3HABITO1HABITO4HABITO5HABITO6SHABIT1SHABIT2
  1. a

    ***p < 0.01, **p < 0.05, *p < 0.1.

Women0.0055−0.0010.0127*0.0228***0.0157**0.0145***−0.0116*0.0197***−0.0068***
Age 30–440.0636***0.015*0.0494***0.0164−0.0311***0.0208**0.00810.0324***0.0067*
Age 45–640.0398***0.01140.0284**0.0317***0.0090.0218**0.0252**0.025***0.0031
Age > 640.01810.0004−0.01160.0272**0.0285**0.0243**0.0396***0.0266***−0.003
Primary school0.044**0.0187−0.0067−0.036**−0.02410.0595***−0.0551***−0.0265*0.0127**
Secondary school0.1017***0.0581***0.0536**−0.0183−0.074***0.0648***−0.0302−0.0275*0.0084
Tertiary school0.1093***0.068***0.0721***−0.0318*−0.111***0.0704***−0.0379*−0.02170.0086
Log income0.1306***0.0386***0.0669***−0.063***−0.0234***0.0597***−0.0144**−0.0117**0.0085***
With dependent children0.00680.0121*0.0108−0.0183**−0.0332***0.0183**−0.0274***−0.007−0.0017
Number of people living in the house0.0056−0.00250.0073*0.0168***0.0132***0.0163***0.012***−0.0028−0.0001
Concern about environment0.032***0.00250.028***0.0234***−0.01220.0188***0.044***0.0574***0.0101***
Knowledge of environmental campaign0.0391***0.024***0.0421***0.021***0.036***0.0472***0.0177**0.0158***0.0076***
Detection of an environmental problem0.0191**0.00850.0421***0.0311***−0.0164*−0.0227***0.0167**−0.0051−0.0029
Organizational collaboration0.0654***0.0546***0.0424**0.0255−0.00480.00680.00420.0036−0.0128
Participating as an environmental volunteer−0.01620.00850.0566**0.0053−0.0089−0.00040.0385*0.0101−0.0059
Signing against environmental damages−0.00430.0272***0.05***0.0264***−0.0246**0.032***0.0384***0.0154**0.0017
Demonstrating against a situation harmful to the environment−0.0424**−0.0094−0.0025−0.00270.0283−0.029*0.00990.00050.0089
Complaint against environmental damages−0.0424**0.0247*0.02990.0414**−0.0552***0.0240.0388**−0.01060.0063
Price−0.00770.0012−0.0102−0.01490.0669***0.0339***0.01520.0184**0.0073
Brand0.0075−0.0271***−0.0384***−0.0233***−0.0116−0.00590.0111*−0.0125***−0.0061**
Efficiency0.049***0.0514***0.0651***0.0194**0.0271***0.0755***0.00130.0137**0.0002
Green−0.00540.00160.01010.0341***0.0365***0.0115*0.0515***0.0092**0.0092***
Local0.00410.0344***0.0355***0.0193***0.0120.0315***−0.0275***−0.0088**0.0031
Penalty for not recycling0.00410.00770.0050.036***0.0026−0.01*0.01160.006−0.0047*
Control for abuse of water use0.0238**−0.00020.0050.0447***0.00680.0415***−0.00720.0373***0.0105***
Environmental tax for polluting fuels−0.0125−0.0002−0.0008−0.0020.0171**0.0445***0.0055−0.00540.0005
Restrictions for private transport0.0044−0.00780.00990.0347***0.0162**−0.00530.0244***0.0070.0032
Ecological tax for tourism0.0168**0.0153***−0.024***0.00820.0171**0.00510.0258***0.0018−0.0019
Renewable energy farm0.0482***0.04***0.0749***0.0239***0.0404***0.0657***0.0254***0.0229***0.007**
Paying more for alternative energy0.0121−0.0119**−0.0179**0.00020.0214**0.0003−0.00070.0060.0007
Decreasing noise on the main streets0.044***0.0136*0.0454***−0.0065−0.00780.0407***0.0060.0133**0.0134***

[46] Among the socioeconomic and demographic variables, the greatest marginal effect is that of a change in income, or those who reach tertiary education, except for the variables, semiclosing the stopcock to reduce the flows, turning the tap off while brushing teeth or lathering, and showering instead of taking a bath. Primary education as the highest qualification has the strongest negative effect on the habit semiclosing the stopcock to reduce flows, whereas it has the strongest positive effect on the habit showering instead of taking a bath. For turning the tap off while brushing teeth or lathering, the greatest marginal effect is for individuals in their 30s or mid-40s.

[47] Among preferences toward the environment, the greatest increases in the probability of household adoption of water-saving equipment come from efficiency being the primary concern in choosing a product, and the organizational collaboration variables. Using full washing and dish-washing machines is associated with efficiency being the primary concern in choosing a product. The variablegreen as the primary concern in choosing a product has the greatest marginal effect, linked to the probability of semiclosing the stopcock to reduce the flows, and price is the main concern for filling the kitchen sink before washing up. Table 9 presents an overview of these results.

Table 9. Overview Multivariate Results
Socioeconomic and Demographic FactorsDISP1DISP2DISP3HABITO1HABITO4HABITO5HABITO6SHABIT1SHABIT2
Women  +++++
Age 30–44+++ + ++
Age 45–64+ ++ +++ 
Age > 64   +++++ 
Primary school+  + ++
Secondary school+++ +  
Tertiary school++++  
Log income+++++
With dependent children + +  
Number of people living in the house  +++++  
Concern about environment+ ++ ++++
Knowledge of environmental campaign+++++++++
Detection of an environmental problem+ +++  
Organizational collaboration+++      
Participating as an environmental volunteer  +   +  
Signing against environmental damages ++++++ 
Demonstrating against a situation harmful to the environment       
Complaint against environmental damages+ + +  
Price    ++ + 
Brand   +
Efficiency++++++ + 
Green   ++++++
Local +++ + 
Penalty for not recycling   +   
Control for abuse of water use+  + + ++
Environmental tax for polluting fuels    ++   
Restrictions for private transport   ++ +  
Ecological tax for tourism++ + +  
Renewable energy farm+++++++++
Paying more for alternative energy  +    
Decreasing noise on the main streets+++ ++ ++

[48] Finally, something must be said about the efficacies of boosting water-saving behaviors. Unfortunately, the microdatabase does not provide information about household water consumption. The INE does provide residential water consumption per capita at the regional level, which allows us, at least, to check whether the annual water consumption growth rate is correlated with WSI, WSEI, and WSHI. The values of WSI, WSEI, and WSHI at the regional level are the arithmetic mean of each region. We find that the correlation between the WSI and the evolution of residential water consumption per capita is −0.2856, more negative again between WSHI and water consumption growth at −0.3933, and weaker between WSEI and water consumption growth at −0.0386. We are aware that this approach is not sufficient to make definitive statements about whether policy makers should give priority to strategies for promulgating water-conservation habits among high-income households, rather than providing subsidies for the purchase of water-saving equipment in low-income households, but it does give us a first intuition. Microdata about both water-saving actions and water consumption is a prerequisite for the implementation of optimal policy measures.

4. Conclusions

[49] This article focuses on household water behavior in Spain, analyzing the influence of a specific set of factors and covering the attitudinal, socioeconomic, and demographic characteristics of Spanish households. Our analysis uses an index developed by the INE that ranks individuals by their water-saving behavior. However, the mixed nature of the data makes it difficult to interpret the quantitative impact of certain independent variables. For this reason, we also analyze the influence of these variables on the household presence of water-saving equipment and, separately, on water-saving habits. Interestingly, we find that, while having both water-saving equipment and water-saving habits leads to water conservation, the influential factors are not the same as those that would explain the ambiguous effects of certain variables found in prior studies. In particular, we find that those individuals most committed to the adoption of water-saving equipment and, at the same time, less committed to water-saving habits tend to have higher incomes.

Appendix A:  

[50] Table A1 presents the descriptive statistics of the variables used in the analysis.

Table A1. Descriptive Statistics
 MeanStd. Dev.MinMax
Dependent variables
 WSI6.3640471.3074371.664510
 WSEI3.3651492.952042010
 WSHI6.2050931.706512010
 DISP10.66970380.470331701
 DISP20.14450780.351612801
 DISP30.30896870.4620801
 HABITO10.20289120.402162401
 HABITO40.38399670.486369501
 HABITO50.81187010.390825801
 HABITO60.28764130.452674501
 SHABIT10.90252470.296611501
 SHABIT20.96442020.185244701
Independent variables
 Women0.5453527 01
 Age51.3015518.686641699
 Age22981.0222004.5632569801
 Age 16–290.1245037 01
 Age 30–440.2945129 01
 Age 45–640.294462 01
 Age > 640.2865214 01
 Years of qualification8.9199334.435052017
 Years of qualification299.2338984.562260289
 Illiterate0.0275883 01
 Primary school0.436272 01
 Secondary school0.3816553 01
 Tertiary school0.1544844 01
 Log income7.2443670.51754076.6200738.116715
 With dependent children0.35819 01
 Number of people living in the house2.5800671.235217113
 Concern about environment0.7704367 01
 Knowledge of environmental campaign0.5968645 01
 Detection of environmental problems0.2555737 01
 Organizational collaboration0.0419424 01
 Voluntary participation0.030337 01
 Sign against environmental damage0.1360073 01
 Demonstration0.0556347 01
 Complaint against environmental damage0.0334419 01
 Price0.9296549 01
 Brand0.5073297 01
 Efficiency0.7732872 01
 Green0.5139469 01
 Local0.5504937 01
 Penalty for not recycling0.5206149 01
 Control for abuse of water use0.7947165 01
 Environmental tax for polluting fuels0.6503614 01
 Restrictions for private transport0.4782144 01
 Ecological tax for tourism0.3407309 01
 Renewable energy farm0.7191286 01
 Paying more for alternative energies0.236384 01
 Decreasing noise in the main streets0.8532526 01

Ancillary