Factors influencing economic benefit of rainwater harvesting: an empirical analysis

The southwestern region of Bangladesh is highly vulnerable to climate change effects and water crises. The Mongla subdistrict of Bagerhat is one of them, where around 65% of people live below the poverty line, and the literacy rate is only 57.20%, while the country's average is 73.91%. Moreover, in this region, 63% of households have access to water year-round, while the country's average is 97 and 88.9% of households use other sources (e.g., rainwater) of water, whereas only 4.4% have tube wells and 6.6% have access to tap water (GoB 2014). Thus, they face triple challenges such as water crises, poverty, and climate-induced extreme events. These challenges are interlinked and force them to live in poor socioeconomic conditions. Therefore, lifting people out of poverty is difficult without mitigating climate change impacts and addressing the water crisis. For this purpose, it needs an economically beneficial and decentralized¹ water supply system (e.g., rainwater harvesting) to address the water crisis problem. In this context, the focal point of this research is to find the factors influencing the economic benefit of a decentralized and alternative water supply system (rainwater harvesting).

Finding the factors that influence the economic benefit of a decentralized water supply system in addressing the water crisis problem is a local, regional, and national level challenge in Bangladesh. Building up the capacity of a reliable water supply system for the community is one of the significant challenges at the local, regional, national, and international levels (Daniel & Tan 2019). For example, many households in Mongla rely on local water traders to supply water, particularly in the dry season. However, water supply from water traders is limited due to their limited financial strength, lack of freshwater sources, climate-induced extreme events, and transportation problems (Islam 2017). In addition, as a coastal area, households face more challenges in receiving a sufficient volume of water through formal and informal channels.

Climate-induced concerns have facilitated the water crisis problem, encouraging new water crisis mitigation with existing resources and technology. As a result, communities, governments, international leaders, and NGOs (Non-Government Organizations) have been looking to address water crisis problems through reliable, economically viable, easy to maintain, and stable water supply systems like rainwater harvesting systems (RHS; Akter & Ahmed 2015).

Rainwater can be collected from the rooftop or in open space through this system. The collected rainwater can be used, stored, and reserved for the future, either underground or above the ground. Gupta et al. (1997) defined rainwater harvesting (RH) as a method for inducing collecting, using, storing, and conserving local surface rainwater for agriculture, while Oweis et al. (2012) found three rainwater harvesting components: catchment, storage facility, and a target. To increase the availability of water for domestic use and agricultural purposes, inhabitants in the arid and semi-arid regions have developed and built different kinds of RHS such as ponds and pans, dams, terracing, percolation, tanks, and Nala bands (Oweis et al. 2012; Adham et al. 2016).

Some academic efforts have been made to measure the value of water, ranging from the willingness to pay, participatory process, and ecosystem services (Kenter et al. 2016), but they are not sufficient and encounter challenges in determining, confirming, and perceiving the entire dimension of social, economic, environmental, and cultural benefits (Garrick et al. 2017) of water. Furthermore, Cai et al. (2006) and Cai & Wang (2006) measured the water resources-economic model and figured out that economic benefits arise from water use in different demand management mechanisms.

Different studies were conducted on economic and financial analysis (Liang & van Dijk 2010; Islam et al. 2014; Matos et al. 2015; Amos et al. 2016), feasibility and measure of climate change adaptation (Pandey et al. 2003; Cook et al. 2013), and sustainability (Rahman et al. 2010; Islam 2017) of rainwater harvesting, but there are limited studies regarding the factors that influence the economic benefit of the same.

Therefore, the discussion concerning cost–benefit analysis and economic feasibility of rainwater harvesting has received attention in the last two decades since the sustainable development issue concerns policymakers, scientists, and civil society. However, although there is greater attention given to finding the economic feasibility of RH for agricultural production, there is no study conducted using household survey/level data to determine the factors that influence economic benefit, but it is more relevant to find out the economic feasibility and possibility to contribute toward poverty reduction, policy formulation, sustainable development, and (rain) water management for socioeconomic development. This study contributes to filling this gap. Furthermore, this study also contributes to poverty reduction and sustainable development goals (SDGs) since rainwater harvesting benefits the household economically. Therefore, this study can play a role in determining factors to promote rainwater harvesting as a tool for poverty reduction, particularly in rain-intensive and water-scarce countries. Moreover, this study will enrich the literature on the economic benefits and feasibility factors of rainwater harvesting, poverty reduction, and SDGs.

This study focuses on whether rainwater runoff from the roofs would be collected and used since roofs are the most applied type of catchment utilized for collecting and using rainwater in the study site. The infrastructure used to collect and use rainwater is known as Rainwater Harvesting System. Figure 1 shows that it has different components such as storage system (runoff accumulates), distribution system (e.g., transfer to a small container to use, transfer to a small or big container to use to conserve), conveyance system or gutter (make a connection between rooftop and storage system), and rooftop from where rainwater collects.

With the unique features of the study site and discussions regarding water as an economic good, common-pool resource, valuation of water, the economic benefit from using water, and pricing of water, the present study focuses on what factors positively or negatively impact total economic benefit arising from water use in households. To answer this question, this paper developed six statistical hypotheses:

Therefore, the sample size is 1,033.

With this sample size determination, data were collected from 1,040 households between August 08 and September 10, 2020, and from July 27 to August 8, 2019, through a structured household survey questionnaire. Data were collected by research assistants (male and female) and the researcher himself. Each of the interviews takes around 25–30 min. A stratified sampling technique was employed to choose a household based on applying a rainwater harvesting system in the household. Every third household is selected in the study site if the household applies rainwater harvesting to address the water crisis. The interpretative variables in Table 1 show the economic, demographic, and social factors that can influence the economic benefit of rainwater harvesting.

This model relates to the components of the modeling framework specified by Ringler et al. (2006). They state that the modeling framework has three components: economic components (e.g., determination of benefits from water use), hydrological components (e.g., reservoirs), and economic incentives and institutional rules that impact the hydrological and economic components. Letcher et al. (2004) developed a similar short- and long-term decision model that attributes policy, economic, and hydrologic components. Islam (2011) argued that this model is based on maximizing net profits (e.g., economic benefit) from residential, agricultural, and industrial water use. In this study, the water storage tank represents the hydrological components, economic benefits represent the economic components and focus on the economic benefit arising from residential water use. Summary statistics are presented in Table 2.

This study is based on cross-sectional survey data and uses a regression model so that the model can provide a biased result that could lead to a less efficient and less reliable model (Hausman 1978). Moreover, Hendry (1980) emphasizes tests to overcome the heteroskedasticity problem. For this reason, this study uses a heteroskedasticity corrected model to strengthen the unbiasedness, increase efficiency, and make the result more reliable. However, to extend our understanding, this study conducts heterogeneity analyses. Firstly, it incorporates socioeconomic attributes into the economic benefit model. Therefore, it could likely influence many factors extending total economic benefit. Secondly, to disentangle the economic benefit component and see whether these are connected to the independent variable. Since the explained variable connects to the explanatory variables, their relationship is expected to go the same way as the primary outcome.

To check robustness, data are divided into two groups: above the median and below the median group, and then we run the model. If the result is the same as the original model in sign and magnitudes, we can assume the model is well-defined and does not have a structural problem. Five robustness analyses were conducted that include: (i) Tot_Eco_Ben²; (ii) total cost³; (iii) storage capacity⁴; (iv) age of rainwater harvesting⁵; and (v) income.⁶

The OLS regression model was employed to test the statistical hypotheses elaborated in Table 3. This method is used to determine the impact of the exogenous variables on the total economic benefit at the household level. A least square regression analysis inaugurated a description of the regression model, measured the model's parameters, and assessed the certainty of the results.

Considering the output in the baseline model in Table 4, all variables used were statistically significant at the 99% confidence level since the p-values were less than 0.001. Moreover, the value of R-squared is 0.632, which entails that the variability of the exogenous variables illustrates 63.20% of the variability of the endogenous variable. Therefore, the number of children negatively and significantly affects the economic benefit. At the same time, storage capacity, age of rainwater harvesting, price, income, and total cost are positively and significantly correlated with the dependent variable. The result in the baseline model indicates that a change in either of the explanatory variables will make a positive change in Tot_Eco_Ben by US$ 112.687 (Tot_Cos_RHS), 100.083 (Age_RHS), 0.574 (Sto_Cap), 1.103 (Inc), 32,708.9 (Pri), and negative change by 35.531 (Num_Chi). This result also indicates how rainwater harvesting can contribute to poverty reduction. For example, a one-liter increase in Sto_Cap will increase Tot_Eco_Ben by 0.574$. This positive economic benefit will positively reflect on the economic strength of the household.

The p-value confirms that all six statistical hypotheses are valid, and the model was correctly specified. The economic benefit and studied factors of rainwater harvesting were significant indicators to determine the total economic benefit arising from rainwater harvesting at the household level. Moreover, the estimated result confirms a statistically significant relationship between economic benefit and household socioeconomic and decentralized water supply systems. Model 2 in Table 4 comprises the result for the heteroskedasticity corrected model. The output of this model does not show any change in the sign for explanatory variables but the change in magnitude compared with the primary model. The Tot_Eco_Ben of rainwater harvesting changes by US$ 1.007 (−), 31.236 (−), 33.321 (+), 0.014 (−), 0.096 (−), and 11,123.5 (−) for Tot_Cos_RHS, Age_RHS, Num_Chi, Sto_Cap, Inc, and Pri, respectively.

Model 3 in Table 4 controls the household's socioeconomic attributes. In this model, the impact of Tot_Cos_RHS on Tot_Eco_Ben rose by 0.021$ while Age_RHS rose by 0.29$. At the same time, the impact of Sto_Cap, Num_Chi, Sto_Cap, and Pri decreased by 0.009$, 8.83$, 0.07$, and 6,022.5$, respectively. Controlling these variables indicates a positive association between Tot_Eco_Ben of rainwater harvesting and age of respondent (0.506), land ownership (1.064) while negative with the household head by male (13.898) and education (5.234) of the respondent. The positive relation of age implies that older people can obtain more economic benefits than young people, while land ownership indicates that households that possess more land can bring more economic benefits to rainwater harvesting. However, there is no change in sign and significant level compared with the primary result.

To determine whether it has heteroskedasticity or not, the Lagrange Multiplier (LM) Breusch–Pagan test was performed. Table 5 presents the Breusch–Pagan test for heteroskedasticity. If it fails to accept the null hypothesis, which indicates that the unit effects’ variance is significantly different from zero, White's heteroskedasticity-consistent standard errors are performed as the solution for heteroskedasticity. This test leaves the OLS coefficients unaffected.

The statistical results of the LM Breusch–Pagan test are presented in Table 4 and show that large LM statistics are associated with a small p-value (below 0.05 or 5%) that indicates we reject the homoscedasticity. Therefore, besides the relationship between explained and explanatory variables, this study conducted a heterogeneity analysis to further understand the relationship between components of economic benefit and factors of economic benefit. Using the original dataset obtained by the household survey to investigate three outcomes of interest: annual economic benefit, annual water collection, and storage capacity to the independent variable.

Model 1 in Table 6 provides the estimated result for annual economic benefit as the dependent variable in which statistically significant factors alter annual economic benefits. These factors are Tot_Cos_RHS (0.179***), Num_Chi (−3.246***), Sto_Cap (0.067***), Inc (0.143***), and Pri (4,010.91***). The regression result also shows that the sign and magnitude of coefficients are similar to the primary outcome in the baseline model in Table 4. The difference is in the Age_RHS, where it is not statistically significant. Secondly, we analyze the annual water collection associated with the independent variable. Annual water collection of 1,040 households relies on storage capacity, the intensity of rainfall, frequency of rainfall, effectiveness in collecting water, availability of people at home during rainfall, and size of the house's rooftop. These are the significant factors that affect the volume of water collection. This water collection affects the economic benefit of rainwater cultivation for fulfilling domestic water demand in the household.

In Table 6, model 2, the OLS coefficients result exhibits that two statistically significant explanatory variables, namely Age_RHS (1,051.13***) and Sto_Cap (8.354***), positively affect annual water collection. The sign of these two variables is matched to the preliminary result obtained but not concerning magnitude. Moreover, the Tot_Cos_RHS and the Num_Chi have a positive correlation but are not statistically significant. In the preliminary result, the number of children is negatively correlated, but here it is positive. It implies that water collection is positively associated with the number of children. Therefore, more children stipulate more annual water collection. However, income and price derive negative coefficients with annual water collection, which is dissimilar to the estimated model.

The regression result obtained in Table 6, model 3, shows several statistically significant variables that include Tot_Cos_RHS (6.270***), Inc (2.162***), Num_Chi (32.083**), and Age _RHS (8.350*) in different confidence intervals with Sto_Cap. The number of children has positive coefficients with storage capacity. It implies that several children in the household urge the family head to enhance the storage capacity. From this point of view, it can assume that households install more water collection tanks with more children in the family.

The model is re-estimated by differentiating between above and below Tot_Eco_Ben. The result (Tot_Eco_Ben) in Table 7 implies that the assumption is valid for Tot_Cos_RHS, Age_RHS, Num_Chi, Sto_Cap, and Pri but not for Inc. More specifically, the Tot_Cos_RHS, Age_RHS, Num_Chi, Sto_Cap, and Pri have more impact by US$0.657, 82.872, 53.696, 0.416, and 32,211.19, respectively, on Tot_Eco_Ben. At the same time, Inc has more impact on Tot_Eco_Ben by 0.052$ in model 2 than model 1 in Table 7 (Tot_Eco_Ben).

The dataset is divided into two sets for the second robustness check: above Tot_Cos_RHS and below-median Tot_Cos_RHS. To the outcome of interest, Table 8 (Tot_Cos_RHS) presents that Age_RHS (45.06$), Num_Chi (38.14$), Sto_Cap (0.072$), Inc (0.994), and Pri (37,443.3$) have more impact on Tot_Eco_Ben for every unit change in above-median total cost household category. Therefore, the assumption is valid for these variables. However, Tot_Cos_RHS (-US$3.412) is less impactful for above-median Tot_Cos_RHS than below-median Tot_Cos_RHS on Tot_Eco_Ben.

Concerning the outcome of interest in storage capacity, Table 8 presents the above and below storage capacity category households. In Table 8, model 1 shows that Tot_Cos_RHS, Age_RHS, Num_Chi, and Pri have better effect by US$ 1.506, 66.319, 71.958, and 49,668.9, respectively, for every unit change in these variable for above-median Sto_Cap. However, the outcomes of Sto_Cap and Inc are reduced by US$ 0.379 and 0.336$ for the same. The output exhibits that assumption is valid for the former group of variables while not for the latter. However, changes in explanatory variables are better described by model 2 since its R-squared value is higher by 17.2% than model 1.

The outcome of interest is Age_RHS. Therefore, the Tot_Cos_RHS, Age_RHS, Sto_Cap, and Inc have a higher impact on Tot_Eco_Ben in model 1 in Table 8 (Age_RHS), while the Num_Chi and Pri have more impact for model 2. More specifically, economic advantages are higher by US$ 0.61, 16.754, 0.394, and 0.904 for every unit alteration through the outcome of the Tot_Cos_RHS, Age_RHS, Sto_Cap, and Inc consecutively for model 1 than model 2 in Table 8 (Age_RHS). Nevertheless, the Num_Chi (66.978$) and Pri (5,534.6$) have a better outcome for model 2 compared with model 1 for each unit change in those variables, respectively.

The estimated result in Table 8 (Inc) by differentiating income groups postulates different contexts and magnitude in the outcome of interest. In Table 8 (Inc), models 1 and 2 are for the above- and below-median Inc group, respectively. The Age_RHS, Num_Chi, Inc, and Pri have better output above-median by US$ 2.919, 17.573, 4.177, and 5,763.8, respectively, for every unit change in these variables. However, the output increases by 0.029$ and 0.04$ in Sto_Cap and Tot_Cos_RHS correspondingly for model 2 compared with the model in Table 8 (Inc). This is attributed to the fact that less economically advantaged people are more efficient in using rainwater to bring economic benefit since their impact of Tot_Cos_RHS and Sto_Cap on the economic benefit is higher than their counterparts. Moreover, their socioeconomic vulnerability may urge them to address their problem themselves, and that may also contribute to more efficiency. As per the result, the presumption is valid for the former group of the variable while not valid for the latter.

This could be attributed to the fact that income is negatively associated with the economic benefit of collecting and using rainwater. The reasons behind this may be a lack of cooperation by the government in addressing the water crisis problem, high-income inequality, the fact that the high-income group may rely on other sources (e.g., bottled water) of water, and governance problems. For example, poor and disadvantaged people should have more access to government support. In practice, they have less access and government support distribution influenced by nepotism. Their income does not improve in this context, but as rainfall is a natural resource, they can easily benefit from it without government intervention.

According to Grafton et al. (2020), decision-makers worldwide face dilemmas in providing reliable water services that ensure basic human needs by maximizing water users’ consumer surplus. They also argued that it is a decision about extending and maintaining water supply and includes economic decisions regarding water pricing to consolidate equity, sustainability, and efficiency. However, Pandey et al. (2003) urge research regarding the economic implications of rainwater harvesting as one of the ways to supply water to households, industry, and agriculture. They state that neither water nor climate policy realizes the worth of rainwater. It needs to be promoted, particularly in water-stressed and rain-intensive countries. This study was carried out in one of the rain-intensive countries and the high-water stress areas. Rainwater collection and use could play an important role in alleviating poverty and achieving SDGs. The rainwater harvesting policy must combine poverty alleviation, economic development, and SDGs. Different countries, even different states, formulate and implement different policies regarding rainwater harvesting.

Although India receives less than half the annual rainfall compared with Bangladesh, every state has a rainwater collection and use policy. For example, rainwater harvesting is mandatory in all commercial and residential buildings in Haryana and Tamil Nadu, irrespective of their size and type. Some states also provide a financial incentive for installing rainwater harvesting. For example, Madhya Pradesh applies a 6% rebate on property tax. However, Holland-Stergar (2015) studied the success and policies adopted in Australia, India, and the United States. The author found that financial investment, the cost to the consumer, and support from NGOs play essential roles in the success and adoption of the policy. The author recommended that policymakers encourage water conservation through rainwater harvesting by providing more financial incentives. They should also consider how the policy will incentivize and employ subsidy or rebate programs (Holland-Stergar 2015).

Policymakers and practitioners must realize the possibility of rainwater to enhance ecosystem productivity, economic development and agricultural production, and reduce vulnerability (Barron 2009). The benefit of rainwater harvesting can end poverty and hunger and ensure gender equality and environmental sustainability as the primary goal of SDGs (Barron 2009). The author urges establishing and enabling policies, providing technical know-how and capacity-building training, and ensuring stakeholder consultation and public participation.

Meehan & Moore (2014) examine the formalization of RHS in the United States. They reveal 96 different rainwater harvesting policies across the country and elaborate on some exciting facts: (i) RHS legislation is relatively recent (2008–2012), updated building and landscape codes that incorporate harvesting of rainwater; (ii) codified rain; (iii) most of the policies regulate micro-level harvesting which is suitable for the single household; and (iv) use markets based tools such as income tax credits, rebates, and other forms of financial incentives to formalize rainwater harvesting policy. The different state uses a different mechanism to formalize policy. For example, the household in California gets US$1.00 per gallon of rainwater collection, while Arizona gets US$2,000.00, which covers half of the material and installation costs (Meehan & Moore 2014). These rebates and incentives are also used in Australia and India to encourage people concerning rainwater use. For example, more than 600,000 Australian households received incentives or government rebates for rainwater harvesting materials like cisterns and barrels in 2010 (Australian Bureau of Statistics 2010), while there was a 6% rebate on property tax in Madhya Pradesh, India.⁷ Moreover, the state (of Brisbane) mandated that newly constructed homes have a rainwater catchment system (Haisman 2005; Moglia et al. 2013).

Therefore, through this decentralized system, legal tools are essential for addressing the water crisis, climate change impact, and drought (Bruch & Troell 2011). However, although some countries have emphasized the importance of rainwater harvesting through formalizing, some still neglect its importance. For example, communities in South Africa have a long history of harvesting rainwater to solve the water crisis, but South African water-connected legislation does not give a concrete legal framework for adopting RHS (Kahinda & Taigbenu 2011). However, Temesgen et al. (2016) observed a lack of clear policy definitions in Ethiopia for rainwater use, while Pacheco et al. (2017) state that Brazilian policies regarding the regulation of rainwater are scattered.

This study has good policy implications regarding the economic benefit of better rainwater management/governance that can contribute to poverty alleviation and sustainable development. This study connects a decentralized water supply system to the economic benefit that can contribute to poverty alleviation and achieve SDGs. Rainwater harvesting contributes to poverty alleviation directly and indirectly. This study reveals the connection between economic benefit and storage capacity, and the cost is favorable. Moreover, the connection between the above-median income and total economic benefit is positive and significant. It implies that higher income means higher economic advantage. Moreover, Islam (2018) found that lack of water negatively influences employment and income and increases household expenditure. These connections mean that the household with higher storage capacity, the cost of rainwater use, and income give a higher economic advantage for using rainwater. We can rightly assume that this economic advantage contributes to the economic development of the household. However, lower-income families have greater efficiency in strengthening economic gain through rainwater harvesting. Since this study connects economic benefits, cost, household income, and price of water, it needs to realize its full potential by accommodating and strictly implementing policy.

Supplementary Tables S1 and S2 list rain-intensive and water-stressed countries. Combining these categories (Supplementary Table S3), several countries evolved which are rain-intensive and face extreme low-level water stress. This study can be the groundwork for identifying how rainwater can perform as an economical means to bring financial advantage to the household. Moreover, it can also lead to a policy implication in reducing water stress and mismanagement of rainwater. Although this study takes place in a low-level water stress country (Bangladesh), it is more applicable to the extreme (e.g., India) or high (e.g., Chile, Nepal, Albania) or medium to high (e.g., Guatemala, Peru, Indonesia, Venezuela, Thailand, Cuba, etc.), or low to medium (e.g., Philippines, Ecuador, Dominica, Haiti, etc.), or low (e.g., Colombia, Bhutan, Guinea, etc.) water stress country with high or medium-intensive rainfall. A highly intensive rain would bring more economic benefits by using rainwater. To find the exact level of economic benefit from a different country, they can initiate a project to find economic feasibility and add the policy to minimize rainwater mismanagement. It will be more important due to climate change, where rainfall will be more erratic, facilitating a water crisis (Islam 2022).

The growing demand for water in entire sectors of the economy, high costs of evolving supplementary supply, and constrained supply have led to competition for current water resources (Booker et al. 2012). Scholars emphasized this issue almost 40 years ago to understand the situation. For example, Randall (1981) featured the situation as moving from an expansionary water economy to a mature water economy where benefits of evolving new water supply go beyond cost, and contemporary supply costs go beyond the limit of benefit, respectively. In this study, rainwater harvesting is a mature water supply source that exceeds the cost.

Several studies on developed and developing economies found that multiple benefits arise from rainwater harvesting on sustainable water use, groundwater recharging, and increasing agricultural production. The positive effect of rainwater harvesting includes the constant increase of economic benefit to the household and increased investment in rainwater harvesting for supplying water for households and agriculture since an economically feasible water supply system is crucial for socioeconomic development.

For the first hypothesis, the positive and significant coefficient evaluation between Tot_Eco_Ben and Tot_Cos_RHS provides a degree of support. Here, total cost acts as an investment in the water supply that contributes to poverty alleviation and socioeconomic development. It incurs a rise in water supply in the household, which positively contributes to economic gain. Moreover, economic gain contributes to alleviating poverty. Investment, water supply, and poverty alleviation are interlinked in this context. The result indicates that increased cost positively contributes to alleviating poverty and economic benefit. The result in Table 5, model 1, characterizes adding 1.627$ to the Tot_Eco_Ben for every unit change in Tot_Cos_RHS. This result is consistent with Haller et al. (2007), Carter & Bevan (2008), and Hanjra & Gichuki (2008), who found that every US$1 investment in water supply provides an economic return between US$ 5 and US$ 45, with the highest return in the least developed countries that can contribute to alleviating poverty. Moreover, this result also enriches the literature by Liang & Dijk (2010), where they found that the cost–benefit ratio is 1.6 (small), 2 (medium), and 2.5 (large). As the ratio in all three sizes is greater than 1, rainwater harvesting is economically viable. Cost is a critical component in measuring the economic feasibility since higher cost needs higher benefits to make it economically sustainable.

Regarding the second hypothesis, the coefficients estimated for the Age_RHS are positively and significantly correlated with Tot_Eco_Ben. This result postulates that increasing the life of a particular rainwater harvesting infrastructure positively contributes to obtaining economic benefit. This condition makes the household use the existing infrastructure as long as possible to increment the benefit. The outcome shows that every year change in the life of rainwater harvesting infrastructure would equal 100.083$ to Tot_Eco_Ben. This result is consistent with Karim et al. (2015), who observed a positive connection between the economic efficiency of rainwater harvesting and greater project life. The finding also develops the argument of Matos et al. (2015), where they found that the Net Present Value (NPV) ranges from $2,66,966.79⁸ to $2,54,667.18 while the Payback Period (PP) is 2–6 years, and IRR (Internal Rate of Return) 23–76%.⁹

The third hypothesis found that the Num_Chi has a negative and significant coefficient with the dependent variable. This negative correlation is attributed to the fact that more family members reduces the total economic benefit. To maintain the economic benefit, it is necessary to increase storage capacity. The OLS result shows that removing a family member could add 35.531$ as an economic benefit from harvesting rainwater. A household with high number of family members needs a bigger storage capacity. It implies that the policymaker needs to consider the family members to decide the contribution to raising the household's water access. The bigger family needs more storage capacity to maintain water access. The output characterizes that providing the same size storage tank to all households irrespective of the number of household members is not effective in addressing the water access problem. Moreover, we can also characterize that number of children reduces the economic capability of the household to invest in rainwater harvesting, so there is a negative relationship between Num_Chi and Tot_Eco_Ben.

The OLS regression result stipulates a positive and significant correlation between Tot_ Eco_Ben and Sto_Cap concerning the fourth hypothesis. This positive relationship means that if the household increases storage capacity, their total economic benefit will increase. The estimated result indicates that Sto_Cap would add 0.574$ to Tot_Eco_Ben for every liter change. This result is consistent with Karim et al. (2015) who found a positive relationship between monetary savings and tank size (storage capacity). However, rainwater harvesting can fulfill domestic water demand from 50 to 90% in Australia (Cook et al. 2013), the United States (Basinger et al. 2010), Taiwan (Lin et al. 2019), and Greece (Londra et al. 2015) with storage capacity between 5 and 76 m³. At the same time, collected rainwater can meet a domestic water demand of 145% in Jordan with a larger storage tank size (Awawdeh et al. 2012). Therefore, it implies a positive relationship between storage capacity and water supply. It is thus assumed that more water supply contributes to economic benefits. In this context, the result is consistent with those studies.

Regarding the fifth hypothesis, the coefficient of OLS regression estimates a positive and significant relationship between economic benefit and price. The estimates suggest that every $ change in price will add 32,708.9$ to Tot_Eco_Ben. This positive connection stipulates that rising prices will positively contribute to the economic benefit of rainwater collection and use. It will encourage the household to use more rainwater instead of contaminated water (e.g., salinity and arsenic) in the surrounding water sources.

The finding is a contrary result for Grafton et al. (2020), who argued that there is a negative relationship between price and benefit. This is consistent with the previous studies of Farreny et al. (2011), Fletcher et al. (2008), and Rahman et al. (2010), who found a positive connection between water price and the economic feasibility of harvesting rainwater. It implies that raising water prices will make rainwater harvesting more economically feasible. In addition, the result enriches the existing literature of Hope et al. (2020), who argued that to make payment for water, there are different payment mechanisms such as monthly fee, pay-as-you-fetch, no payment, pay-when-break, and pay as per demand. This study follows the last approach (pay as per demand). The finding also develops the pricing mechanism of water. For example, Beecher (2020) proposed an all-inclusive pricing model that attributes design cost-based rates for different use of water (resource management), acknowledges public functionality in cost allocation (scope economics), limits the service rather than disconnection (water security), provides a basic use allowance for entire households (public health), and determines a minimum bill to property evaluation (capacity value).

Regarding the sixth hypothesis, the household income level positively and significantly correlates with economic benefits arising from rainwater collection and use. It implies that the more income the household has, the more economic benefit from rainwater use. It exposes the connection between investment capacity in rainwater harvesting and economic benefit. We can assume that higher-income families have a more substantial financial capacity to invest in the water supply system. More investment brings more economic benefits. It elaborates that the wealthier household has more financial strength to extend the economic benefit of rainwater harvesting. This result is consistent with Sadoff et al. (2015: 45), who observed that ‘water-related investments can increase economic productivity and growth, and economic growth can provide the resources to finance capital-intensive investments in water-related infrastructure’. This result is consistent with Foster & Hope (2016) and Hope et al. (2020), who revealed that households who used water for economic activities such as small-scale irrigation and livestock watering would make more regular payments for water since these activities generate income and make them affordable to pay.

The most critical perspective of applying the economic benefit model is obvious household income and how rainwater harvesting contributes to that income directly and indirectly. The study hypothesized that hydrology factors have a strong and positive impact on total economic benefit at the household level, which was confirmed by the result. The result in Table 5, model 1, stipulates that every unit change in Tot_Cos_RHS, Age_RHS, Num_Chi, Sto_Cap, Inc, and Pri will impact US$ 1.627, 100.083, −35.531, 0.574, 1.103, and 32,708.9, respectively, on Tot_Eco_Ben.

Moreover, the estimated result in Table 4, model 3, with adding demographic and socioeconomic attributes shows land (1.064) and age (0.506) have positive, while education (−5.234) and male household head (13.898) have a negative correlations with Tot_Eco_Ben. This indicates that education and male household head are factors in reducing Tot_Eco_Ben while land and age are rising. From here, we can conclude that women are more efficient at bringing economic benefit through rainwater harvesting, managing rainwater harvesting infrastructure, and utilizing collected rainwater. Although the heteroskedasticity corrected model (model 2 in Table 4) reveals less impact of each variable on Tot_Eco_Ben, there is no significant change in terms of magnitude and sign.

Model 2 in Table 6 found that Num_Chi has a positive association (44.548) with annual water collection. This result indicates that more children means more water collected through rainwater harvesting. Since Tot_Eco_Ben has a positive connection with annual water collection, more children can contribute to more economic benefits from rainwater harvesting. In this context, in one aspect, more children are a burden for the household, but assets when it comes to poverty reduction since they can enhance economic benefit through rainwater harvesting. However, model 3 in Table 7 also presents that Num_Chi positively correlates with Sto_Cap. It stipulates that the household head installs more storage tanks for rainwater harvesting as family members increase. Policymakers should consider this aspect to promote rainwater harvesting to improve water access, especially in coastal areas.

The result from robust analysis suggests the above-median household group has a higher effect by the explanatory variable on economic benefit, except for one or two variables for every robust analysis. For example, Inc (0.052) for Tot_Eco_Ben, Tot_Cos_RHS (3.412) for Tot_Cos_RHS, Sto_Cap (0.379) and Inc (0.336) for Sto_Cap, Num_Chi (66.978) and Pri (5,534.6) for Age_RHS, Tot_Cos_RHS (0.029) and Sto_Cap (0.04) for Inc group. This implies that the assumption is valid for four or five variables in each analysis. This result indicates the robustness of the finding.

This study urges five issues that can be considered to address the water access problem, minimize mismanagement of (rain) water, achieve SDGs, and alleviate poverty. Firstly, rainwater harvesting can be regarded as a tool to help poverty reduction strategy. According to the result in Table 5, there is a positive correlation between storage capacity and economic benefit so that policymakers can emphasize how to enhance the storage capacity of the household to increase the water supply. This improvement of the water supply will reduce the cost of water for that household and increase the economic benefit. Secondly, this study recommends preparing a rainwater harvesting policy like India since the country (e.g., Bangladesh) faces severe challenges in managing heavy rainfall during monsoon (which causes floods almost every year) and minimizing mismanagement of (rain)water. Moreover, it is one of the most rain-intensive countries, and this abundant rainfall flows to the sea with damaging water management infrastructure. Thirdly, the other related policies (e.g., National Water Management Plan, Bangladesh Water Act, Poverty Reduction Strategy, etc.) need to update and recognize rainwater harvesting as a tool for better (rain) water management and governance in addressing the water crisis and poverty alleviation and achieving SDGs.

Fourthly, policies and programs supporting rainwater harvesting (e.g., microcredit loans targeting small farmers and financially disadvantaged groups, financial incentives for installing rainwater harvesting infrastructure and using rainwater, and making rainwater harvesting mandatory for residential and commercial buildings irrespective of size and location) have proved an effective tool that should be promoted. Finally, this study urges further study to determine the adequate size of the storage tank with due consideration of the number of family members (adult and child).

This research could be helpful for local, regional, and national public authorities involved in improving access to water for the household, increasing household income, poverty alleviation, and implementing water management policy. Moreover, it is also helpful for the scientific communities interested in the interconnection between water, rainwater harvesting, economic benefit, poverty alleviation, and household income. This study may also be supplemented by extending the analyzed geographical frame, collecting data by fieldwork, and using different statistical methods. To assess the factors that interact with the economic benefit of rainwater harvesting, the innovation of this research nests in the novelty outlook taken for a set of 1,040 relevant households affected by the water crisis problem. The previous study focused on the economic benefit of rainwater harvesting from irrigation and hydrology.

Although the study provides an insightful understanding of the relationship between economic benefit, poverty reduction, rainwater harvesting, and sustainable development, it has some limitations. Firstly, this study takes place in a middle rain-intensive country and in the rural/coastal areas. Secondly, the number of observations. Thirdly, time and budget. Further research is needed for more conclusive findings by overcoming these limitations.

This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 754345.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.