The Impact of Rural Water Supply and Sanitation on Economic Growth in South Asia

On the other hand, it was found that rural access to improved sanitation Granger causes GDP growth (although not homogenously across countries). It showed that there[r]

(1)

1 The Impact of Rural Water Supply and Sanitation on Economic

Growth in South Asia

M G C N Madadeniya University of Peradeniya

Sri Lanka Abstract

"Clean water and sanitation" is one of the 17 Sustainable Development Goals set by the United Nations, and the achievement of this goal is crucial for rural South Asia which accounts for around one third of the world’s rural population Clean water and sanitation facilities reinforce economic growth by creating a healthy labour force, increasing environmental sustainability and supporting economic activities in all sectors of an economy This study aims to analyse the impact of rural people’s access to clean water and sanitation facilities on economic growth in South Asia, as the lack of scientific studies focusing on the particular issue for South Asia is a gap in existing literature A panel data econometric analysis was conducted including four South Asian countries, for the period 1991-2015 The findings of the panel regression indicated that the rural population’s access to improved water sources and improved sanitation facilities has no significant impact on the region’s economic growth during the period studied However, Granger causality analysis showed that rural population’s access to improved sanitation facilities Granger causes economic growth Further, it revealed that rural population’s access to improved water sources as well as capital growth Granger cause rural population’s access to improved sanitation facilities Given that capital growth had a significant positive impact on economic growth in the regression analysis, more capital investments in the rural water supply and sanitation projects are thus encouraged for the South Asian region to reap growth benefits which still remain undiscovered Although labour force growth had a negative, significant impact on economic growth in the regression analysis, causality analysis revealed that growth in labour force Granger causes rural people’s access to improved sanitation facilities homogenously across countries Hence, technological developments and other investments in human capital can help improve the productivity of labour and thereby facilitate its contribution to economic growth in South Asia

(2)

2 1 Introduction

South Asia is home to around one third of world’s rural population In fact, around 66 percent of the South Asian population live in rural areas of the region Therefore, rural development is essential for South Asia to achieve higher economic growth In this context, the provision of clean water and sanitation facilities to the rural communities can be considered as an important strategy to achieve rural development

However, the supply of water and sanitation facilities to rural South Asia is inadequate In 2015, around 86 percent of rural population in South Asia used at least basic drinking water facilities and only 37 percent used at least basic sanitation facilities During the same year, only 54 percent of rural population in the region used safely managed drinking water facilities and only 26 percent used safely managed sanitation facilities

However, as shown by Figure 1, rural people’s access to improved water sources and sanitation facilities has gradually increased throughout the past

0 20 40 60 80 100

1990 1995 2000 2005 2010 2015

Improved water source, rural (% of rural population with access) Improved sanitation facilities, rural (% of rural population with access)

Year %

Figure 1: Percentage of Rural People with Access to Improved Water and Sanitation Facilities in South Asia (1990-2015)

Source: World Development Indicators

(3)

3 Access to these improved water and sanitation facilities can improve health and nutrition level of rural people, support productive household and commercial activities, save time spent by rural women on carrying water from distant sources to household premises, increase quality of life, ensure environmental sustainability and finally create a healthy and productive rural labour force which is conducive to economic growth

In 1990, only 66 percent of the rural South Asian population had access to improved water sources and only percent of them had access to improved sanitation facilities But in 2015, 91 percent of the rural South Asian population had access to improved water sources and 35 percent of them had access to improved sanitation facilities

Hence, the problem statement of the study can be specified as follows

As the majority of the South Asian population lives in the rural areas of the region, their access to improved water and sanitation facilities can have massive economic benefits During the period from 1990 to 2015, the South Asian economy grew at an average rate of 6.2 percent As a matter of fact, South Asia is the fastest growing region in the world today In this context, the problem arises whether rural water supply and sanitation in South Asia, which have increased over the past, have an impact on the region’s economic growth The particular issue motivated the study to address the following questions

 Do rural water supply and sanitation induce economic growth in South Asia?  Can South Asian countries use rural water supply and rural sanitation as strategies

to achieve higher economic growth?

The main objective of this study is to identify the impact of rural water supply and sanitation on economic growth in South Asia As specific objectives, it attempts to study the relationship between rural water supply, rural sanitation and economic growth in South Asia, and also to draw policy implications of the findings

Clean water and sanitation is the 6th sustainable development goal in the 2030 Agenda for sustainable development The universal access to clean water and sanitation facilities is a key foundation to achieve not one, but many sustainable development goals such as, no poverty, zero hunger, good health and well-being, decent work and economic growth etc The achievement of this goal in South Asia is very important as it is one of the most populated regions in the world today In this context, this study provides a guideline for the policy makers, emphasizing the impact of rural water supply and sanitation on economic growth in South Asia, and showing how the current situation can be improved for higher economic growth in the future

(4)

4 Section includes the results and discussion Finally, section gives the conclusions and policy implications of the study

2 Literature Review

Many of the previous studies have identified a strong link between rural development and economic growth in South Asia These studies are noteworthy when beginning a discussion on the impact of rural water supply and sanitation on economic growth in the region

Naseem (2004) in the book on rural development and poverty in South Asia, have mentioned that for the predominantly agricultural economies of South Asia, rural development is the core issue of development Further, it is stated that due to the low priority given to rural areas in the national development, South Asian rural societies have suffered a steady erosion in their living conditions and productive infrastructure, as evidenced by the high incidence of poverty Khan (2015) has studied the nexus between rural development, growth and poverty reduction in South Asia In the particular study, it is stated that the development experience of South Asia suggests a strong link between rural development, growth and poverty reduction It is emphasized that from growth and poverty reduction perspective, rural development must be given priority in the development process of the region The study identifies low access to safe drinking water and proper sanitation facilities as an important indicator which reflects rural poverty in South Asia World Bank (2016) reveals that South Asia accounts for one third of the global poor, 80% of which live in rural areas of the region Therefore, rural development can significantly enhance economic growth by increasing the economic contribution of these people

Many studies are available in the literature that focus on the impact of water and sanitation on economic growth and development in various regions and countries in the world

Okun (1988) has analysed the value of water supply and sanitation programmes for economic development and listed out their benefits which are: preventing diseases; improving primary health care; improving nutritional status; facilitating health centres, clinics and schools; saving time spent on fetching water; facilitating household irrigation and animal watering; promoting commercial activities; supporting all economic sectors; strengthening community organization; and improving quality of life Thus, the particular research supports the view that water and sanitation increases economic growth

(5)

5 limited relative to current and future populations, countries may find it difficult to generate additional growth through more water use It is concluded that privatization, pricing reforms and water markets can establish incentives for more efficient use of water in the economy than the water management solely by public sector Thus, Barbier shows the possibility for a negative water-growth relationship in highly populated regions like South Asia

Nevertheless, most of the studies emphasize water and sanitation as essential factors contributing to economic growth Anderson and Hagos (2008) have explored the potential linkages between access to water and sanitation and growth-related indicators in Ethiopia using an econometric analysis Results show that the improvements in the source of drinking water are likely to report an improvement in household food situation No significant relationship has been found between improvements in drinking water sources and households’ overall welfare Changes in sanitation arrangements also have shown no relationship with changes in households’ food or welfare situations

Hutton et al (2008) have examined the economic impacts associated with poor sanitation in Cambodia, Indonesia, the Philippines and Vietnam They have found that poor sanitation causes considerable financial and economic losses in the four countries The major contributors to overall economic losses are the cost of premature death (mainly of children under five years old), time spent accessing unimproved water and sanitation facilities, tourism losses, health care costs, sickness time etc

Musouwir (2010) has conducted a time series analysis and found a statistically significant relationship between national budget on water supply and sanitation and GDP per capita in 22 African countries The study strongly encourages governments of developing countries to spend more of their annual budgets on the water sector

Minh and Huang (2011) have conducted a study on the economic impacts of unimproved sanitation in developing countries Their evidence prove that the economic cost associated with poor sanitation is more than the economic cost associated with poor water supply Previous studies have also attempted to derive numerical estimates of the costs and benefits associated with water and sanitation OECD (2011) has shown that benefits from the provision of basic water supply and sanitation services are massive and far outstrip costs It says, benefit-to-cost ratios have been reported to be as high as to for basic water and sanitation services in developing countries In the meantime, their findings reveal that there may be some “disbenefits” in providing access to water, sanitation and hygiene, depending on the sequencing of investments, for example, if access to water is provided without simultaneous access to sanitation However, adequate water and sanitation services are identified as key drivers of economic growth

(6)

6 Productivity losses due to absenteeism and high medication costs due to various health issues are identified as some economic impacts

According to Frontier Economics (2012) providing universal access to water for all poorly-serviced populations worldwide will cost at least USD 175 billion An additional USD 550 billion would be required to provide universal access to sanitation services In India, one-off investment requirements are $4,338 million and $36,911 million for water and sanitation respectively Annual potential economic gain is $16,550 million In Nepal, one-off investment requirements are $142 million and $896 million for water and sanitation respectively Annual potential economic gain is $389 million In Pakistan, one-off investment requirements are $965 million and $3852 million for water and sanitation respectively Annual potential economic gain is $1454 million Although the initial investment requirements are large, if properly maintained, they can create cumulated benefits with time

UNU and UNOSD (2013) have found that the inadequate water and sanitation infrastructure increases expenditures in other sectors Costs will be incurred when getting water from informal private providers, health costs increase because of waterborne diseases, and the potential for human productivity is also compromised

Almost all of these studies highlight the fact that lack of access to water and sanitation incurs excessive costs to an economy which exceed the costs that should be borne when providing them However, Ibok and Daniel (2014) have pointed out that along with public spending on water supply and sanitation, some other conditions should also be met to achieve favourable economic outcomes Analysing the rural water supply in Akwa Ibom State in Nigeria, they have identified the lack of maintenance, lack of community participation, lack of coordination and co-operation among the stakeholders, political factors, inefficient monitoring, and poor attitude towards public property etc., as reasons for unproductive government expenditures on water supply It emphasizes that continuous maintenance, coordination and regulation are needed to make the water supply projects conducive to economic growth

Some more recent studies further prove the importance of water supply and sanitation for an economy Patunru (2015) has illustrated the importance of water and sanitation facilities by estimating their impact on diarrhoea incidence in Indonesia The study finds that when it comes to health issues, the relative importance of sanitation is higher than that of water Sadoff et al (2015) have showed that South Asia has the largest global concentration of water-related risks, with the largest global concentration of people without adequate sanitation and growing environmental threats India, with its very large population, is the top-ranked country globally for the number of people without adequate water supply and sanitation This raises the question whether the inadequacy of water supply and sanitation in South Asia has any unfavourable impact on economic growth in the region

(7)

7 be employed in eight water and natural resource-dependent industries: agriculture, forestry, fisheries, energy, resource intensive manufacturing, recycling, building and transport UN (n p) has shown that by managing water sustainably, production of food and energy can be better managed and it will also contribute to decent work and economic growth It is also mentioned that although the extension of basic water and sanitation services to the unserved will cost a significant amount, the costs are even higher if poor water supply and poor sanitation are left unsolved According to the World Bank estimates, 6.4 percent of India’s GDP is lost due to adverse economic impacts and costs of inadequate sanitation OECD (n p) has shown that the greatest economic losses from water insecurity are resulted by inadequate water supply and sanitation, associated loss of life, health costs, lost time, and other opportunity costs

Through the above literature review, it was clear that water supply and sanitation play an important role in the growth of an economy However, relatively less attention has been given to the impact of water supply and sanitation on economic growth in South Asia Moreover, the supply of water and sanitation facilities to rural South Asia and its impact on economic growth have hardly been subjected to scientific studies

3 Methodology

This study conducted a panel data analysis including data for four South Asian countries which are, Bangladesh, India, Pakistan and Sri Lanka Annual data were collected from the World Development Indicators database of the World Bank and the sample period was from 1991 to 2015 The four countries as well as the sample period were selected according to the availability of data A balanced panel was used in this analysis where each country has the same number of observations

The Neo-classical growth accounting equation was followed in constructing the model Growth accounting explains what part of growth in total output is due to growth in different factors of production Output grows through increases in inputs and also through increases in productivity The production function expresses the quantitative relationship between inputs and outputs Assuming labour (N) and capital (K) as the only inputs, the following equation shows that output (Y) depends on inputs and the level of technology (A)

(1)

The following growth accounting equation is derived from the above function

(2) Y = A 𝑓(K, N)

∆Y Y = θ

∆N

N + (1 − θ) ∆K

K + ∆A

(8)

8 where,

θ = labour’s share of income 1-θ = capital’s share of income

Thus, labour and capital, each contributes an amount equal to their individual growth rates multiplied by the relevant input’s share of total income The last term of the equation, ∆A/A, is the rate of improvement in technology Hence, output growth can be expressed as a linear function of labour growth, capital growth and technological growth The value of ∆A/A is often described as an estimate of total factor productivity (TFP) growth or the Solow residual It can be calculated from the above equation as a residual, by subtracting capital growth and labour growth (multiplied by the relevant input’s share of total income) from output growth However, according to Barro (1998), an alternative approach would be to regress the growth rate of output on the growth rates of inputs, so that the intercept will measure the value of ∆A/A, and the coefficients of the factor growth rates will measure the capital’s and labour’s shares of income Although with certain limitations, this alternative approach provides a simple way of decomposing growth rate of output into components associated with factor accumulation and technological progress

Following the alternative approach suggested by Barro (1998), an econometric model was constructed, incorporating the target variables into the growth equation The model is specified below

(3)

where,

GDPG = Growth of Gross Domestic Product (at constant 2010 US $) GCFG = Growth of Gross Capital Formation (at constant 2010 US $) LFG = Growth of labour force

LNW = Log of the number of rural people with access to improved water sources LNS = Log of the number of rural people with access to improved sanitation facilities t = time or year

The methodological tools used in this study were as follows i Panel Unit Root Tests

Three panel unit root tests were conducted to check the stationarity property of the data series They are; Levin, Lin & Chu t test, Augmented Dickey Fuller-Fisher Chi-Square test and Philips-Perron-Fisher Chi-Square test

ii Panel Regression

In order to find the relationship between the dependent variable and independent variables, Panel regression methods were used According to Zulfikar (2018), estimating the regression

(9)

9 model using panel data can be done through three approaches, among others: Pooled Least Squares regression model (Common Effect model), Fixed Effects model and Random Effects model The study estimated all these three models

 Pooled LS Regression model

In Pooled Least Squares regression model, all observations are simply pooled and regression analysis is carried out, ignoring the cross-section and time series nature of the data, in which case the error term captures everything This camouflages the heterogeneity or individuality that exists between the variables

 Fixed Effects model

The Fixed Effects model differs from the Pooled Least Squares regression, but still uses the ordinary least square principle Nwakuya and Ijomah (2017) states that the fixed-effects model controls for all time-invariant differences between the individual entities In other words, it controls for the average differences across groups (countries) in any observable or unobservable predictors Thus, it assumes the same slopes and constant variance across countries In the basic Fixed Effects model, the effect of each predictor variable (i.e., the slope) is assumed to be identical across all the groups, and the regression merely reports the average within-group effect Thus, the estimated coefficients of the fixed-effects models cannot be biased because of omitted time-invariant characteristics such as culture, religion etc So Fixed Effects models are considered a method to avoid the problem of omitted variable bias

As Fixed Effects model controls for all the across-group action, what remains is the within-group action In fact, fixed-effects models are designed to study the causes of changes within a group, assuming that the time invariant characteristics are unique to the group and should not be correlated with other groups’ characteristics Therefore, one country’s error term and the constant (which captures individual characteristics) should not be correlated with the others Hence, for a Fixed Effects model, a reasonable amount of variation of independent variables is needed within each group However, the effect of the other important variables that have little within-group variation cannot be assessed in this method

Alternatively, estimating Fixed Effects model panel data using dummy variables, i.e Least Squares Dummy Variable (LSDV) method, can capture the differences among groups Here, it is assumed that differences between groups can be accommodated from different intercept, given that the independent variables are non-stochastic

 Random Effects model

(10)

10 unobservable factors are assumed to be time-invariant, then Fixed Effects regression will eliminate omitted variable bias If the unobservable factors are not time-invariant, i.e if they move up and down over time within categories in a way that is correlated with the variables included in the regression, then the omitted variable bias is still present In such cases Random Effects model should be estimated But this model is also vulnerable to omitted variable bias (omitted-variable bias occurs when a statistical model leaves out one or more relevant variables, resulting in the attribution of the effect of the missing variables to the estimated effects of the included variables)

Random Effects model does not use the principle of ordinary least square, but uses the principle of Maximum Likelihood or Generalized Least Square (GLS) The difference among groups (or time periods) lies in their variance of the error term, not in their intercepts In other words, residuals may be correlated between time and between groups An advantage of Random Effects model is that the estimates can be generalized as the model assumes that it analyses a sample taken from a population Conversely, in Fixed Effects model, the estimates are specific to the data used In addition, Random Effects model also eliminates heteroscedasticity

To decide the best model out of the three above regression models, three types of tests were conducted

iii Hausman test

The choice between Fixed Effects model and Random Effects model depends on the presence of individual heterogeneity among groups, which can be tested using Hausman Test If the null hypothesis is rejected, then it is concluded that there is individual heterogeneity and therefore, Random Effects model is not appropriate, alternatively, Fixed Effects model should be used

iv Redundant Fixed Effects test

(11)

11 v Lagrange Multiplier (LM) test

LM test helps to decide between a random effects regression and a simple OLS regression The null hypothesis in the LM test is that variances across entities is zero, i.e no significant difference or heterogeneity across groups (no panel effect) Four LM tests were conducted in this study which are, Breusch-Pagan, Honda, King-Wu, Standardized Honda and Standardized King-Wu

Causality tests were also conducted to find the causal relationship between variables in the model

vi Granger Causality test

In line with most of the literatures in econometrics, one variable is said to Granger cause the other if it helps to make a more accurate prediction of the other variable than had we only used the past of the latter as predictor Granger causality between two variables cannot be interpreted as a real causal relationship but merely shows that one variable can help to predict the other one better (Awe 2012: 2)

Two types of Granger Causality tests were conducted which are: Stacked Causality Test and Dumitrescu-Hurlin Causality test The stacked causality test treats the panel data set as one large stacked set of data without taking a lagged value of one cross section to the next cross section This approach assumes that all coefficients are same across all groups (common coefficient) Dumitrescu-Hurlin causality test allows for all coefficients to be different or heterogeneous across groups This approach takes into account two different statistics which are:Wbar-statistic (takes average of the test statistics) and Zbar-statistic (shows a standard, asymptotic normal distribution)

4 Results and Discussion

First, panel unit root tests were conducted to test the stationarity property of the time series1 Three unit root tests were conducted, namely; Levin, Lin & Chu t test, Augmented Dickey Fuller-Fisher Chi-Square test and Philips-Perron Fisher Chi-Square test Levin, Lin & Chu t test assumes common unit root process in the null hypothesis whereas all the other unit root tests assume an individual unit root process in the null hypothesis The following table summarizes the results of the unit root tests

(12)

12

Table 1: Panel unit root test results

Levin, Lin & Chu t ADF – Fisher Chi-square PP – Fisher Chi-Square test

Variable Level Level Level

GDPG -5.608***

(0.000)

39.943*** (0.000)

44.498*** (0.000)

GCFG -7.310***

(0.000)

61.900*** (0.000)

64.363*** (0.000)

LFG -1.771**

(0.038)

29.240*** (0.000)

29.060*** (0.000)

LNW -5.881***

(0.000)

27.244*** (0.001)

114.270*** (0.000)

LNS -13.674***

(0.000)

130.609 (0.000)

56.326*** (0.000) Note: *, ** and *** indicate rejection of null hypothesis at 10%, 5% and 1% respectively

Accordingly, at 5% level of significance, all the variables in the model were stationary at level Therefore, it was clear that Panel Least Squares regression methods can be adopted to find the relationship between variables

(13)

13

Table 2: Pooled LS Regression results

Variable Coefficient Std Error t-Statistic Prob

GCFG 0.156*** 0.018 8.695 0.000

LFG -0.241*** 0.090 -2.681 0.009

LNW 0.105 0.183 0.573 0.568

LNS 0.281 0.247 1.136 0.259

C -2.081 2.321 -0.897 0.372

R-squared 0.541 Mean dependent var 5.272

Adjusted R-squared 0.522 S.D dependent var 1.928

S.E of regression 1.333 Akaike info criterion 3.462

Sum squared resid 168.848 Schwarz criterion 3.592

Log likelihood -168.085 Hannan-Quinn criter 3.514

F-statistic 28.007 Durbin-Watson stat 1.821

Prob(F-statistic) 0.000

Note: *, ** and *** indicate rejection of null hypothesis at 10%, 5% and 1% respectively

The growth in gross capital formation and labour force have a significant impact on GDP growth at percent level of significance Holding other factors constant, when growth in gross capital formation increases by one percent, GDP growth increases by 0.156 percent on average Holding other factors constant, when growth in labour force increases by one percent, GDP growth decreases by 0.281 percent on average Rural people’s access to improved water and sanitation not have a significant impact on GDP growth

(14)

14 -20 -10 10 20 30 40 B G D B G D B G D B G D B G D B G D IN D 92 IN D 95 IN D 98 IN D 01 IN D 04 IN D 07 IN D 10 IN D 13 P A K 92 P A K 95 P A K 98 P A K 01 P A K 04 P A K 07 P A K 10 P A K 13 LK A LK A LK A LK A LK A LK A

Growth in capital to labour ratio Growth in GDP

Growth in gross capital formation Growth in labour force

Country Year %

Figure 2: Growth in GDP, capital, labour and capital to labour ratio across the four South Asian countries: Bangladesh, India, Pakistan and Sri Lanka

Source: World Development Indicators

The model’s R-squared value which is equal to 0.541 indicates that around 54.1 percent of the total variability of GDP growth is simultaneously explained by the independent variables of the model The adjusted R-squared is 52.2 percent It shows that the independent variables can explain 52.2 percent of the variability in GDP growth

According to the individual t tests, all other variables except for capital growth and labour growth are insignificant variables However, individually insignificant variables can be jointly significant in explaining economic growth In such a case overall significance of the model can be tested using F test where the null hypothesis is that the independent variables of the model jointly have no significant impact on the dependent variable

(15)

15 improved water and sanitation facilities could have a significant impact on economic growth jointly with other variables in the model

Next the Fixed Effects model (LSDV) and Random Effects model were estimated

Table 3: Fixed Effects Model results (cross sectional Fixed Effects)

GCFG 0.151*** 0.018 8.318 0.000

LFG -0.130 0.108 -1.211 0.229

LNW 2.950** 1.458 2.023 0.046

LNS -0.449 0.448 -1.002 0.319

C -41.484** 20.581 -2.016 0.047

Effects Specification

Cross-section fixed (dummy variables)

S.E of regression 1.318 Akaike info criterion 3.467

Sum squared resid 159.886 Schwarz criterion 3.676

Log likelihood -165.359 Hannan-Quinn criter 3.552

The growth in gross capital formation and rural people’s access to improved water facilities have a significant impact on GDP growth at percent level of significance Holding other factors constant, when growth of gross capital formation increases by one percent, GDP growth increases by 0.151 percent on average Holding other factors constant, when rural people’s access to improved water facilities increases by one percent, GDP growth increases by 2.95 percent on average All the other variables not have a significant impact on GDP growth

(16)

16 The model’s R-squared value and adjusted R-squared value are 56.5 percent and 53.2 percent respectively, which are higher than those of pooled regression It indicates that the ability of the independent variables to explain the total variability of GDP growth simultaneously is higher in Fixed Effects model

The following table shows the results of Random Effects model

Table 4: Random Effects Model results

GCFG 0.152*** 0.0181 8.428 0.000

LFG -0.155 0.106 -1.460 0.148

LNW 1.495 1.027 1.456 0.149

LNS -0.102 0.374 -0.274 0.785

C -20.927 14.573 -1.436 0.154

Effects Specification

S.D Rho

Cross-section random 3.403 0.870

Idiosyncratic random 1.318 0.131

Weighted Statistics

S.E of regression 1.311 Sum squared resid 163.342

Unweighted Statistics

R-squared -0.214 Mean dependent var 5.272

Sum squared resid 446.840 Durbin-Watson stat 0.674 Note: *, ** and *** indicate rejection of null hypothesis at 10%, 5% and 1% respectively

Only the growth in gross capital formation has a significant impact on GDP growth at percent level of significance Holding other factors constant, when growth of gross capital formation increases by one percent, GDP growth increases by 0.152 percent on average

(17)

17 water and sanitation facilities could have a significant impact on economic growth jointly with other variables in the model

The model’s R-squared value and adjusted R-squared value are 47.0 percent and 44.8 percent respectively, which are lower than those of pooled regression and Fixed Effects model It indicates that the ability of the independent variables to explain the total variability of GDP growth simultaneously is lower in Random Effects model

In this scenario, it was important to select the most appropriate model which is able to analyse the impact of rural water supply and sanitation on economic growth correctly Hausman test was used to select the most suitable model out of the Fixed Effects and Random Effects models The results are given below

Table 5: Hausman Test results

Test Summary Chi-Sq Statistic Chi-Sq d.f Prob

Cross-section random 1.988 0.738

According to the results, the null hypothesis that Random Effects model should be chosen, could not be rejected It indicated that the existence of unobservable and time-variant factors, which are correlated with the variables included in the regression, is possible

Further proof was needed before the elimination of the Fixed Effects model from the analysis Therefore, Redundant Fixed Effects test was conducted to check if fixed effects are necessary for this panel-regression The results of the test are given below

Table 6: Redundant Fixed Effects test results

Effects Test Statistic d.f Prob

Cross-section F 1.932 (3,68) 0.133

Period F 1.172 (24,68) 0.298

Cross-Section/Period F 1.242 (27,68) 0.234

(18)

18 The p-values associated with the F-statistic and the Chi-square statistic are both greater than 0.05 Therefore, the null hypothesis cannot be rejected at percent level of significance This suggested that there is no unobserved heterogeneity among countries Thus, fixed effects are redundant and therefore unnecessary

After eliminating Fixed Effects model from the analysis, five Lagrange Multiplier (LM) tests were conducted to determine whether Random Effects model is better than Pooled LS method The results are given below

Table 7: LM Tests results

Test Hypothesis

Cross-section Time Both

Breusch-Pagan 1.260 0.037 1.297

(0.262) (0.848) (0.254)

Honda -1.123 0.192 -0.657

(0.869) (0.424) (0.744)

King-Wu -1.123 0.192 -0.994

(0.869) (0.423) (0.840)

Standardized Honda -0.386 0.305 -4.930

(0.650) (0.380) (1.000)

Standardized King-Wu -0.386 0.305 -4.861

(0.650) (0.380) (1.000)

The null hypothesis of LM tests assumes that Pooled LS method is better than Random Effects model As shown above, the results of LM tests indicated that the null hypothesis cannot be rejected Thus, agreeing with the Redundant Fixed Effects test, it too showed that there is no unobserved heterogeneity among countries Hence, most appropriate model for this study was the Pooled LS method

Finaly, Pair-wise Granger Causality tests were conducted2 The following table summarizes the results

(19)

19

Table 8: Pair-wise Granger Causality test results

Variable Stacked test Dumitrescu Hurlin test

Growth in gross capital formation

Granger causes rural access to improved sanitation

Does not homogenously Granger cause rural access to improved sanitation Growth in labour force Does not Granger cause

rural access to improved water sources

Homogenously Granger causes rural access to improved water sources Granger causes rural access

to improved sanitation

Homogenously Granger causes rural access to improved sanitation Rural people’s access to

improved water sources

Granger causes rural access to improved sanitation

Homogenously Granger causes rural access to improved sanitation Rural people’s access to

improved sanitation facilities

Granger causes GDP growth Does not homogenously Granger causes rural access to improved water sources Granger causes rural access

to improved water sources

Homogenously Granger causes rural access to improved water sources GDP growth Granger causes gross capital

formation growth

Homogenously Granger causes gross capital formation growth Does not Granger cause

labour force growth

Homogenously Granger causes labour force growth Granger causes rural access

to improved sanitation

Homogenously Granger causes rural access to improved sanitation

5 Conclusions

(20)

20 Results showed that the rural population’s access to improved water sources as well as rural population’s access to improved sanitation facilities not have a significant impact on economic growth in the selected four South Asian countries during the period studied However, the Granger Causality tests proved that it they can still contribute to economic growth in South Asia

Capital growth had a positive and significant impact on economic growth, and it was also found that capital growth Granger causes rural access to improved sanitation (although not homogenously across countries) On the other hand, it was found that rural access to improved sanitation Granger causes GDP growth (although not homogenously across countries) It showed that there is a potential impact of rural people’s access to improved sanitation facilities on economic growth Moreover, it was found that rural people’s access to improved water sources Granger causes rural access to improved sanitation facilities and rural people’s access to improved sanitation facilities Granger causes rural access to improved sanitation facilities homogenously across countries Therefore, more capital investments in the rural water supply and sanitation projects are encouraged for the South Asian region to reap growth benefits which still remain undiscovered

Although labour force growth had a negative and significant impact on economic growth, causality analysis revealed that labour force growth Granger causes rural access to improved sanitation homogenously across countries It means that more employment can lead to more access to improved sanitation facilities However, policy makers should ensure the productivity of an increasing labour force Technological developments and other investments in human capital can help improve labour productivity and thereby facilitate its contribution to economic growth

Another interesting finding was that GDP growth Granger causes capital growth, labour growth as well as rural people’s access to improved sanitation facilities homogenously across countries This shows that the effective implementation of other economic policies that aim higher economic growth will help increase the level of rural sanitation in South Asia Policies aiming higher capital growth is more important because capital growth can improve both economic growth and rural sanitation in South Asia

(21)

21 6 References

Anderson, E and Hagos, F (2008) ‘Economic impacts of access to water and sanitation in Ethiopia: Evidence from the welfare monitoring surveys’ Working Paper No Department of International Development [Online] https://www.gov.uk/dfid-research-outputs/working- paper-3-economic-impacts-of-access-to-water-and-sanitation-in-ethiopia-evidence-from-the-welfare-monitoring-surveys Accessed 30 October 2018

Awe, O.O (2012) ‘On Pairwise Granger Causality Modelling and Econometric Analysis of Selected Economic Indicators’ Interstatt journals [Online]

interstat.statjournals.net/YEAR/2012/articles/1208002.pdf Accessed 30 October 2018 Barbier, E.B (2004) ‘Water and economic growth’ Economic Record Vol 80, no 248, pp 1-16

Barro, R.J (1999) ‘Notes on Growth Accounting’ Journal of Economic Growth Vol 4, no 2, pp 119-37

Frontier Economics (2012) Exploring the links between water and economic growth London, Frontier Economics Ltd [Online]

https://freshwaterwatch.thewaterhub.org/sites/default/files/final-frontier-report_update18092012_0.pdf Accessed 30 October 2018

Hutton, G., Rodriguez, U E., Napitupulu, L., Thang, P and Kov, P (2007) Economic impacts

of sanitation in Southeast Asia Washington DC: World Bank [Online]

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1149071 Accessed 30 October 2018 Ibok, E.E and Daniel, E.E (2014) ‘Rural Water Supply and Sustainable Development in Nigeria: A Case Analysis of Akwa Ibom State’ American Journal of Rural Development Vol 2, no 4, pp 68-73

Khan, M.A (2017) Nexus between Rural Development, Growth and Poverty Reduction in

South Asia [Online]

https://www.researchgate.net/publication/265261694_Nexus_between_Rural_Developme nt_Growth_and_Poverty_Reduction_in_South_Asia_1 Accessed 30 October 2018

Minh, H V and Hung N V (2011) ‘Economic aspects of sanitation in developing countries’

Environmental health insights no 5, pp.63-70

Musouwir, T H (2010) ‘Water and economic development: correlation between investment in the water sector and economic growth in developing countries’, Doctoral dissertation, UNESCO-IHE

Naseem, S M (2004) Rural Development and Poverty in South Asia: Development Papers

(22)

22 Nwakuya, M.T and Ijomah, M.A (2017) ‘Fixed Effect Versus Random Effects Modeling in a Panel Data Analysis; A Consideration of Economic and Political Indicators in Six African Countries’ International Journal of Statistics and Applications Vol 7, no 6, pp 275-279 OECD (2011) Benefits of Investing in Water and Sanitation [Online]

http://www.oecd.org/berlin/47630216.pdf Accessed 30 October 2018 OECD (n p.) Water, growth and finance: Policy perspectives [Online]

https://www.oecd.org/environment/resources/Water-Growth-and-Finance-policy-perspectives.pdf Accessed 30 October 2018

Okun, D.A (1988) ‘The value of water supply and sanitation in development: an assessment’ American Journal of Public Health Vol 78, no 11, pp 1463-1467

Patunru, A.A (2015) ‘Access to safe drinking water and sanitation in Indonesia’ Asia & the

Pacific Policy Studies Vol 2, no 2, pp 234-244

Sadoff, C W., Hall, J W., Grey, D., Aerts, J C J H., Ait-Kadi, M., Brown, C., Cox, A., Dadson, S., Garrick, D., Kelman, J., McCornick, P., Ringler, C., Rosegrant, M., Whittington, D and Wiberg, D (2015) Securing Water, Sustaining Growth: Report of the GWP/OECD Task Force

on Water Security and Sustainable Growth United Kingdom: University of Oxford [Online]

https://www.gwp.org/globalassets/global/about-gwp/publications/the-global-dialogue/securing-water-sustaining-growth.pdf Accessed 30 October 2018

Tezera, B.S (2011) ‘Water supply and sanitation: development impacts of poor accessibility of potable water supply and basic sanitation in rural Ethiopia: a case study of Soddo District’, Master's thesis, University of Agder

UN (2016) Water and jobs World Water Assessment Programme Secretariat (WWAP) Paris, United Nations Educational, Scientific and Cultural Organization [Online]

unesdoc.unesco.org/images/0024/002439/243938e.pdf Accessed 30 October 2018 UN (n p.) Clean water and sanitation: why it matters [Online]

http://www.un.org/sustainabledevelopment/water-and-sanitation/ Accessed 30 October 2018

UNU and UNOSD (2013) Water for Sustainability: Framing Water within the Post-2015

Development Agenda United Nations University [Online]

https://www.sei-international.org/ /unosd-unu-sei-watersdg-full-report.pdf Accessed 30 October 2018 World Bank (2016) Poverty and Shared Prosperity 2016: Taking on Inequality Washington, DC: World Bank Group [Online]

https://openknowledge.worldbank.org/handle/10986/25078 Accessed 30 October 2018 Zulfikar, R (2018) ‘Estimation Model And Selection Method Of Panel Data Regression : An Overview Of Common Effect, Fixed Effect, And Random Effect Model’ [Online]

(23)

23

Appendix

i Panel Unit Root Test Results

 Real GDP growth (GDPG)

Panel unit root test: Summary Series: GDPG

Date: 07/05/19 Time: 14:12 Sample: 1991 2015

Exogenous variables: Individual effects Automatic selection of maximum lags

Automatic lag length selection based on SIC: to

Newey-West automatic bandwidth selection and Bartlett kernel

Cross-

Method Statistic Prob.** sections Obs

Null: Unit root (assumes common unit root process)

Levin, Lin & Chu t* -5.60833 0.0000 92

Null: Unit root (assumes individual unit root process)

Im, Pesaran and Shin W-stat -5.13949 0.0000 92 ADF - Fisher Chi-square 39.9427 0.0000 92 PP - Fisher Chi-square 44.4975 0.0000 96 ** Probabilities for Fisher tests are computed using an asymptotic Chi

-square distribution All other tests assume asymptotic normality

 Real Gross Capital Formation Growth (GCFG)

Panel unit root test: Summary Series: GCFG

Date: 07/05/19 Time: 14:15 Sample: 1991 2015

Automatic lag length selection based on SIC:

Newey-West automatic bandwidth selection and Bartlett kernel Balanced observations for each test

Cross-

(24)

24 Levin, Lin & Chu t* -7.31038 0.0000 96

Im, Pesaran and Shin W-stat -7.85943 0.0000 96 ADF - Fisher Chi-square 61.8998 0.0000 96 PP - Fisher Chi-square 64.3630 0.0000 96

** Probabilities for Fisher tests are computed using an asymptotic Chi -square distribution All other tests assume asymptotic normality

 Labour Force Growth (LFG) Panel unit root test: Summary Series: LFG

Date: 07/05/19 Time: 14:19 Sample: 1991 2015

Exogenous variables: None

Automatic selection of maximum lags

Newey-West automatic bandwidth selection and Bartlett kernel Cross-

Levin, Lin & Chu t* -1.77112 0.0383 94

ADF - Fisher Chi-square 29.2396 0.0003 94 PP - Fisher Chi-square 29.0594 0.0003 96 ** Probabilities for Fisher tests are computed using an asymptotic Chi -square distribution All other tests assume asymptotic normality

 Log of the number of rural people with access to improved water sources (LNW) Panel unit root test: Summary

Series: LNW

Date: 07/05/19 Time: 14:17 Sample: 1991 2015

(25)

25 Automatic selection of maximum lags

Newey-West automatic bandwidth selection and Bartlett kernel

Cross-

Levin, Lin & Chu t* -5.88142 0.0000 93

Im, Pesaran and Shin W-stat -1.85686 0.0317 93 ADF - Fisher Chi-square 27.2442 0.0006 93

PP - Fisher Chi-square 114.270 0.0000 96

** Probabilities for Fisher tests are computed using an asymptotic Chi -square distribution All other tests assume asymptotic normality

 Log of the number of rural people with access to improved sanitation facilities (LNS) Panel unit root test: Summary

Series: LNS

Date: 07/05/19 Time: 14:18 Sample: 1991 2015

Newey-West automatic bandwidth selection and Bartlett kernel Cross-

Levin, Lin & Chu t* -13.6739 0.0000 92

Im, Pesaran and Shin W-stat -18.8107 0.0000 92 ADF - Fisher Chi-square 130.609 0.0000 92

PP - Fisher Chi-square 56.3261 0.0000 96

(26)

26

ii Granger Causality test results

 Stacked Causality Test results Pairwise Granger Causality Tests Date: 07/05/19 Time: 14:26 Sample: 1991 2015

Lags:

Null Hypothesis: Obs F-Statistic Prob

GCFG does not Granger cause GDPG 96 0.01596 0.8997

GDPG does not Granger cause GCFG 4.18228 0.0437

LFG does not Granger cause GDPG 96 1.30653 0.2560

GDPG does not Granger cause LFG 1.34603 0.2489

LNW does not Granger cause GDPG 96 2.65771 0.1064

GDPG does not Granger cause LNW 0.74075 0.3916

LNS does not Granger cause GDPG 96 5.94089 0.0167

GDPG does not Granger cause LNS 8.05118 0.0056

LFG does not Granger cause GCFG 96 0.19499 0.6598

GCFG does not Granger cause LFG 1.96293 0.1645

LNW does not Granger cause GCFG 96 0.71625 0.3995

GCFG does not Granger cause LNW 0.00244 0.9607

LNS does not Granger cause GCFG 96 2.44363 0.1214

GCFG does not Granger cause LNS 3.91483 0.0508

LNW does not Granger cause LFG 96 1.19329 0.2775

LFG does not Granger cause LNW 0.01000 0.9206

LNS does not Granger cause LFG 96 0.04851 0.8262

LFG does not Granger cause LNS 10.7190 0.0015

LNS does not Granger cause LNW 96 18.1679 5.E-05

(27)

27

 Dumitrescu-Hurlin causality test results

Pairwise Dumitrescu Hurlin Panel Causality Tests Date: 07/05/19 Time: 14:27

Sample: 1991 2015 Lags:

Null Hypothesis: W-Stat Zbar-Stat Prob

GCFG does not homogeneously cause GDPG 0.51536 -0.69589 0.4865 GDPG does not homogeneously cause GCFG 3.76210 3.13418 0.0017

LFG does not homogeneously cause GDPG 1.60137 0.58524 0.5584 GDPG does not homogeneously cause LFG 2.94271 2.16757 0.0302

LNW does not homogeneously cause GDPG 1.49504 0.45980 0.6457 GDPG does not homogeneously cause LNW 1.15068 0.05358 0.9573

LNS does not homogeneously cause GDPG 1.52970 0.50069 0.6166 GDPG does not homogeneously cause LNS 3.05193 2.29642 0.0217

LFG does not homogeneously cause GCFG 0.83229 -0.32202 0.7474 GCFG does not homogeneously cause LFG 0.64013 -0.54870 0.5832

LNW does not homogeneously cause GCFG 0.30217 -0.94738 0.3434 GCFG does not homogeneously cause LNW 0.94261 -0.19187 0.8478

LNS does not homogeneously cause GCFG 0.28967 -0.96212 0.3360 GCFG does not homogeneously cause LNS 1.12889 0.02788 0.9778

(28)

28 LNS does not homogeneously cause LFG 2.05437 1.11963 0.2629 LFG does not homogeneously cause LNS 10.0514 10.5535 0.0000