Do the Improved Water Sources for 203 Countries Converge over Time?

Yusang Chang,Yootaek Lee,Yoonji Lee,Ki Bae Kim

1 Gachon University

2 Questrom School of Business,Boston University

3 Frederick S.Pardee School of Global Studies,Boston University

4 Maxwell School of Citizenship and Public Affairs,Syracuse University

Keywords Improved water source Convergences analysis Sigma index Reduction of dispersion Gamma index Catch-up effect

Abstract The Sustainable Development Goals by the United Nations include a global target for Improved Water Source to reach 100% of population by 2030.There are still millions of people lacking the improved access to drinking water. We use the convergence methodology for 203 countries to determine whether the water shortage gap between water-rich versus water-poor countries has narrowed during 2000 to 2015. Results of our analysis show the narrowing of water shortage gaps. However,significant variations are found among income and regional subgroups. Implications of our findings indicate the reduction of water shortage gap within appropriate subgroups is needed. In particular,a set of policy recommendations are presented for sub Saharan Africa region where improvement of water supply is urgently needed.

1 Introduction

Access to safe drinking water is critical for promoting human health,socioeconomic development,and individual wellbeing. (Hsu et al., 2016). The Global Sustainable Development Goals (SDGs) established by the United Nations in 2015 include a global target for safely managed drinking water supply. (United Nations,2016). SDG target 6.1 stated that by 2030, achieve universal and equitable access to safe and affordable access to drinking water for all. These new goals align with the United Nations’formal acknowledgement that clean drinking water and sanitation are encompassed in the realization of human rights(United Nations,2010).

According to the report from Joint Monitoring Program (JMP) by UNICEF/World Health Organization(2015),more than 91%of the global population had access to improved water source for their drinking water by 2015. The annual measure used is officially known as Improved Water Source (IWS)which refers to the percentage of the population using safely managed drinking water services which is located on premises,available when needed and compliant with faecal and priority chemical standards (UNICEF/World Health Organization,2017).

Table 1 Comparison of IWS scores between 5 lowest and 5 highest countries sub Saharan Africa(2000-2015).

However, large access gaps in improved water source exist between developing and developed regions.Developed regions have made substantial progress gaining access to improved water source, but developing regions still have many peoples with limited access to improved water source. For example,the averaged IWS in 2015 for the Sub Saharan Africa region was only 72.68%. In contrast,the high income group of 64 countries achieved 98.69%by 2015. In addition to 319 million people in Sub-Saharan Africa,61 million people in South-Eastern Asia, 134 million people in Southern Asia and, 65 million people in Eastern Asia are still waiting for the improved water source for drinking in 2015(JMP,2015).

Within a given subgroup of countries, access gaps also exist among individual countries, sometimes due to differences in population density or income distribution. For example,the lowest IWS scores for the 5 countries in 2015 from the Sub Saharan African Region were Somalia(25.8%),Equatorial Guinea(47.9%),Angola(49%),Chad(50.8%)and Mozambique(51.1%). The average score of these 5 countries was 44.92%,as shown in Table 1. On the other hand,the near perfect IWS scores for the 5 countries,excluding those countries with the perfect score of 100% in 2015 were Mauritius (99.9%), Sao Tome (97.1%), Botswana (96.2%), Seychelles (95.7%),and South Africa (93.2%). Their averaged 2015 score was 96.42%, representing a very wide water access gap compared to the 5 lowest IWS group,both of which were within the same region of Sub Saharan Africa.

Whenever such a wide water access gap exists between the two groups of countries,the issue of convergence becomes quite relevant (Bhattacharya et al., 2018; Huang et al., 2018; Yu and Zhang, 2015; Zumaquero and Rivero,2016). If the countries with poor IWSs improve faster than the countries with superior IWSs,the former may gradually catch-up to the latter for a convergence. For example,the averaged IWS scores for the 5 countries with the best IWS score increased from 90.86%in 2000 to 96.42%in 2015 at the Compounded Annual Growth Rate(CAGR)of 0.42%. On the other hand,the CAGR for the group of 5 lowest IWS score was much higher at 0.70% increasing from the 2000 score of 40.46% to 44.92%, as shown in Table 1. The net effect was that the group of 5 lowest IWS score was slowly catching up to the group of 5 highest IWS score over time. In short,this example represents the case of an increasing averaged IWS scores with a slowly decreasing country differences of IWS measure or a decreasing dispersion of IWS scores for the combined group of 10 countries. In short,the convergence between the two groups of 5 countries appear to suggest that convergence analysis when applied to global samples of multiple countries may yield similar results.

The general convergence question of this research is whether country differences of IWS measures existing in 2000 have decreased by 2015? If so, how fast is the speed of decrease? More specifically, we ask whether countries with initially low IWSs improve faster than countries with initially high IWSs? If so, how fast is the speed of catch-up. Second,we ask whether dispersion of IWSs for multiple countries decreased over time.If so, how rapid is such reduction of dispersion. These two specific convergence questions will be analyzed for the total group of 203 countries as well as for the four subgroup of countries by income levels and six subgroups of countries by regions. To our best knowledge,such broad-based convergence studies on IWSs have not been published with one exception(Neumayer,2001). Therefore,this research may make a contribution to the literature on convergence analysis of water shortage. On the other hand,it should be noted that this study is not designed to explore various endogenous and exogenous factors affecting IWS dynamics.

After this introduction, description of convergence methods we use will be presented in the second section.Data and data sources will follow in the third section. The fourth section presents the results of analysis. Finally,conclusions and limitations of this study will be presented in the fifth section.

2 Methodology

This The initial idea of convergence (also known as catch-up effect) is based on the hypothesis that poorer countries’ economy has a tendency to grow faster than richer countries so that the former can catch-up to the latter. Conventionally, the term ‘convergence’ has two connotations in economic growth literatures. First, the term can refer to a reduction of dispersion among countries involved,which is known asσconvergence. Second,the term can also mean the phenomenon that poorer countries grow faster than richer ones, which is known asβconvergence. In the context of convergence analysis of IWS measures, countries with initially lower IWSs are likely to improved their IWSs faster than countries with initially higher IWSs over time for a catch-up. And then,country differences of IWSs will diminish resulting in a reduction of dispersion among countries involved.

Theβconvergence method(Barro,1991)regress the rate of change by comparing beginning year value with ending year value of the performance measure for respective countries. When the slope of regression equation is negative and statistically significant, the convergence is confirmed. (Barro, 1991; Barro and Sala-i-Martin,1992). Such approach is known as absoluteβconvergence in which all countries are assumed to move toward a common destination. However, the use of regression for convergence analysis received critical objection by Friedman (1992) citing the so-called Regression Fallacy explaining the natural tendency of regression towards the mean. Friedman instead recommended another method for convergence analysis, which is known asσconvergence. He asserted that convergence pattern could be more properly measured by tracing the fluctuation of coefficient of variations of performance measures for given countries. If the trend of fluctuation is declining and statistically significant,σconvergence can be demonstrated.

Quah(1996)also criticized the use ofβconvergence analysis because it did not explain the inter-temporal change or intra-distribution of mobility among countries involved. Quah(1993)instead suggested using Markov Chain analysis by which a researcher can track the dynamics of cross-country distribution. Similar to Quah’s approach, a simple measure ofβconvergence was proposed by Boyle and McCarthy (1997). They used rank concordance of Kendall’s index (Siegel, 1956) to measure chronological changes in the ranking of countries,which is calledγconvergence. They suggested that the combination ofσconvergence andγconvergence could be proper alternative to theβconvergence because the combination could not only tell us the existence and its speed of catch-up effect but also capture the dynamics of distribution of countries involved.

Since then,a large number of studies usingγconvergence methodology have been published in such areas as energy, economic growth,inflation,employment, and healthcare (Agovino and Rapposelli,2017; Bhattacharya et al., 2018; Carrasco and Ferreiro, 2014; Ferrara and Nistico, 2015; Huang et al., 2018; Jaunky and Zhang,2016;Kallioras and Tsiapa,2015;Yu and Zhang,2015;Zumaquero and Rivero,2016;Chang et al.,2019a).

For our research, we have adopted the method ofγconvergence (Boyle and McCarthy, 1997) andσconvergence(Friedman,1992)to analyze convergence among 203 countries. Generally,the standard deviation and coefficient of variation are used as parameter of dispersion (Heckelman, 2015). Particularly for our research,we have decided to use coefficient of variation (CV)forσconvergence analysis. CV is measured by dividing standard deviation by the sample average. The inter-temporal changes could be measured by normalizing a CV of subsequent year to the initial CV. Hence, normalized CV of beginning year is always 1. If the CVs in the subsequent years are less than the CV in the initial year,the normalized CV in subsequent years will be less than 1. When the normalized CVs in the subsequent years keep declining and the differences between CV of initial year and that of subsequent years are statistically significant,σconvergence can be confirmed. For statistical test of difference, we used sample t-test for CVs(http://www.real-statistics.com/students-t-distribution/coefficientof-variation-testing/). The test works well if the sample sizes are more than or equal to 10. Since the sample sizes in our research are much larger than 10,the test should be effective.

As mentioned above,Boyle and McCarthy(1997)proposed the use of rank concordance which measures individual countries’mobility over time within the cross-country distribution(Liddle,2012;Chang et al.,2019b).To put it another way,γconvergence quantifies the degree of ordinal ranking change of countries between the initial year and a given year. Theγconvergence method has two kinds: Binary Kendall method and Multi-annual Kendall Method. We decided to use Binary Kendall method for our analysis. The method is defined as follows:

Where,AR(γ)= the actual rank of countryi’s performance measure, in yeart;AR(0) =the actual rank of countryi’s performance measure,in year 0;γt=Binary KendallγIndex in yeart.

Analogous with the normalized CV forσconvergence analysis, theγindex has an important advantage of tracing the degree of change over time. The index can range from zero to unity. If there is no change in the ranking order, the index becomes unity. If a catch-up effect exists, the actual rank of a country in year t will change which results in reduction of nominator, accordingly, the index will show the value less than unity. The test statistic is chi-square which is used to test whetherγindexes show any significant difference between ranks of beginning year and given year(Siegel,1956).

As the website of Real Statistics Using Excel explains (http://www.real-statistics.com/reliability/kendallsw/),the requirement is that the number of countries involved should be more than or equal to 5,or the number of years being compared should be more than or equal to 15. In our research, the sample size should be much larger than 5. Therefore,we can usex2test to validate the null hypothesis stated above.

How do we useσandγindex together to evaluate reduction of dispersion as well as catch-up process?There are four different cases that can occur. The simplest case is when bothσandγindex are increasing in values. Under the circumstance, neither reduction of dispersion nor catch-up may be taking place. The second case is that bothσandγindexes are decreasing which indicates that both reduction of dispersion and catch-up process are taking place. The third case occurs whereσconvergence measure is non-decreasing,whileγconvergence value is in decline. Sinceβconvergence is a necessary but not sufficient condition forσconvergence,this indicates that catch-up process is taking place,while reduction of dispersion is not. The fourth case occurs whereγindex is non-decreasing while a substantial decline occurred withσindex. This indicates that country differences in performance measures remain so that no rank change among countries takes place.However, performance differences among countries have reduced considerably, which indicates conditionalβconvergence. Put it another way,catch-up process may be taking place within subgroups of countries.

3 Data and data sources

For this study,we have used Improved Water Source(IWS)data from WHO/UNICEF Joint Monitoring Program for Water Supply and Sanitation. IWS for a given country is measured as percentage of population which fulfilled access to improved water resource for their drinking. We downloaded yearly IWS data for 233 countries and territories around the world during 2000-2015 from the World Bank website(https://data.worldbank.org/indicator/SH.H2O.SAFE.ZS).

From this list, 20 entities with missing yearly IWS data between 2000 to 2015 were eliminated. Thus, we ended up with 213 countries and territories. The list of 20 deleted entities are as follows: They are: 1.Bermuda 2.Brunei Darussalam 3.Curacao 4.Faeroe Islands 5.Falkland Islands 6.Gibraltar 7.Holy See 8.Hong Kong 9.Isle of Man 10.Liechtenstein 11.Macao 12.Mayotte 13.Norfolk Island 14.Pitcairn 15.Saint Helena 16.Saint Pierre and Miquelon 17.San Marino 18.Saint Martin 19.Svalbard and Jan Mayen Islands 20.Wallis and Futuna.

For categorizing income subgroups,we have used the World Bank’s GNI per capita data(https://data.worldbank.org/indicator/NY.GNP.PCAP.CD). According to the World Bank, GNI per capita is the gross national income, converted to U.S. dollars using the World Bank Atlas method, divided by the midyear population. According to the World Bank, four income groups are categorized in 2015 as follows. The high income group contains those countries whose GNI per capita of$12,746 or more,followed by the upper middle income group with GNI per capita between $4,126 and $12,745. The lower middle income group contains those countries between$1,045 and$4,125,while the lower income group contains those countries with less than$1,045.

In matching these 213 countries and territories to the income data available at the World Bank Site,we find the following 10 entities whose income data are not listed at the web site. The list of 10 missing entities are as follows: They are: 1.Anguilla 2.Cook Islands 3.French Guiana 4.Guadeloupe 5.Martinique 6.Montserrat 7.Niue 8.Reunion 9.Taiwan 10.Tokelau. After deleting these 10 entities,we ended up with the final sample size of 203 countries for our convergence analysis.

To categorize regional subgroups,we have also referenced country classification of World Bank(https://datahelpdesk.worldbank.org/knowledgebase/articles/906519worldbank-country-and-lending-groups). According to World Bank,the total countries can be divided by 7 regions of East Asia and Pacific,Europe and Central Asia,Latin America &the Caribbean, Middle East and North Africa, North America, South Asia, and Sub-Saharan Africa. For our analysis, we added North America region to Europe and Central Asia region since only 3 countries are included in the North America, following the same combined regional categorization used by UNICEF/WHO report(2017).

Table 2 presents historical averaged IWSs for the total group of 203 countries as well as for the four subgroups of countries with income levels of high,upper middle,lower middle,and low following the World Bank categorization. The IWS for the total group of 203 countries has improved from 83.07% in 2000 to 88.65%in 2015 at the compounded annual growth rate (CAGR)of 0.43%. The high income subgroup of 64 countries shows the lowest CAGR of 0.06%due to the high initial IWS of 97.78%in 2000 which has improved slightly to 98.69%by 2015. The low income group of 31 countries has the highest CAGR of 1.21%which has increased its 2000 IWS of 57.70% to 69.14%by 2015. The low middle income group of 52 countries shows the second highest CAGR of 0.79%,whereas the upper middle income group of 56 countries has the CAGR of 0.32%. The 2015 IWSs are 93.75% for the upper middle income group and 82.41% for the lower middle income group.In short, the higher the income level, the higher the averaged IWSs and the lower becomes the annual rate of improvement in IWSs.

Table 3 shows historical averaged IWSs for the total group of 203 countries as well as for the 6 regional subgroups of East Asia and Pacific (EAP), Europe, Central Asia and North America (ECNA),Latin America and Caribbean (LAC), Middle East North Africa (MENA),South Asia (SA),and Sub-Saharan African (SSA)regions following the categorization by the World Bank.

The most rapid annual rate of increase at 1.05%was realized by the 48 countries in SSA which recorded the lowest IWS of 62.16%in 2000. The 8 countries in SA closely followed with 1.04%which recorded the second lowest IWS of 76.41%in 2000. In contrast,the 53 countries in ECNA realized the second lowest annual rate of increase at 0.14%together with the 21 countries in MENA with the lowest annual rate of increase at 0.09%. Both ECNA and MENA recorded high 2000 IWSs at 95.32%and 89.43%respectively. The two remaining regions of EAP with 34 countries and LAC with 39 countries experienced moderate annual rate of increase at 0.51%and 0.32%respectively. Their 2000 IWSs were 89.49%for LAC and 83.76% for EAP.In sum,the annual rates of increase by regions, in general, displayed the similar pattern shown earlier by income subgroups. Namely, the higher the 2000 IWS,the slower becomes the annual rate of increase of IWS during the period of 2000 to 2015.

Table 2 Averaged IWSs for total and four income subgroups of countries(2000-2015).

Table 3 Averaged IWSs for total and six regional subgroups of countries(2000-2015).

Table 4 Normalized Sigma and Gamma indexes of total group of 203 countries(2000-2015).

4 Analysis of results

Normalized yearlyσandγindexes for the total group of 203 countries are shown in Table 4. Bothσandγindexes display statistically significant declining patterns indicating that convergence of IWS has taken place.However,the rate of decline between the two indexes was quite different. Normalizedσindex has declined from 1.0 at the beginning year of 2000 to 0.758 by 2015 at the negative compounded annual growth rate(CAGR)of-1.83%. Statistical significant differences were observed each year during the period of 2006-2015. On the other hand,normalizedγindex has declined more slowly from 1.0 in 2000 to 0.905 by 2015 at the negative CAGR of-0.66%. Statistically significant differences were verified between each year during the whole period between 2001 to 2015 at less than the 1%level.

It was indicated earlier that the averaged IWS has increased from 83.07%in 2000 to 88.65%by 2015 for 203 countries as shown in Table 1. The results fromσandγconvergence provide some important additional insights for the total group of 203 countries. The results fromσconvergence indicate that dispersion of IWS for the total group measured in yearly CV has declined by 24.2%. In other words,differences of IWS for individual countries have decreased substantially suggesting that the gap between countries with higher IWS versus countries with lower IWS has narrowed. The results fromγconvergence indicate a moderate catch-up effect as the normalizedγindex of 1.0 has decreased by 9.5% during this period. This means that a number of countries with lower IWS have overtaken some other countries with higher IWS by achieving more rapid improvement. In sum,both reduction of dispersion fromσconvergence and catch-up effect fromγconvergence indicate that a substantial narrowing of country differences of IWS has accompanied the improved averaged IWS for the total groups of 203 countries during 2000 to 2015.

In addition, we divided the total group of 203 countries into the four subgroups of countries according to their income level of high, upper middle, lower middle, and low and shows the results in Table 5. As forσindexes, the fastest convergence has taken place in the high income subgroup of 64 countries at the CAGR of

-2.21%, which suggests that as more nations of high income approach 100% access to drinking water, rapid reduction of dispersion has taken place. On the other hand, the slowest convergence has occurred in the upper middle income subgroup at the CAGR of-0.64%. Lower middle and low income subgroups realized somewhat more rapid declining rates of -1.51% and -1.68%, which are closer to the declining rate for the total group. It should be noted that the results of statistical tests were not significant for these subgroups with an exception of the high income subgroup.

Table 5 Normalized Sigma and Gamma indexes of four income subgroups(2000-2015).

***Significant at 1%level,**Significant at 5%level,*Significant at 10%level

As forγindexes, statistical tests generate significant differences of allγindexes covering each year during the period as shown in Table 5. The higher was the income level, the faster was the speed of reduction ofγindexes. The most rapid reduction rate of-1.96%for the high income subgroup was followed by-1.35%by the upper middle,-0.94%by the lower middle and finally-0.8%for the low income subgroup.

It is interesting to note that the sequence of income subgroups on the speed of catch-up reversed the annual improvement rate of IWS measures. In other words, the higher the income level, the slower was the annual improvement rate of IWS.However,the speed ofγconvergence becomes faster. One possible explanation may be that the higher income subgroups are more likely to have smaller country differences in IWS measures. For example, 2015 averaged IWS measures for the high and upper middle income subgroups were 98.69% and 93.75%respectively as opposed to 69.14%by the low income subgroup. Under the circumstances, the chances of ranking changes among individual countries would be much higher in the higher income subgroups.

Table 5 shows the distribution of normalizedσandγindexes for 6 regional subgroups. Inσindexes the most prominent feature was a divergence in MENA region where the 2015σindex increased to 1.1914 from 1.0 in 2000 at the annual speed of+1.17%. On the other hand,all the remaining 5 regions displayed decliningσindexes at the speed ranging from-3.1%by SA to-1.07%by LAC.However,a majority ofσindexes did not meet the statistical test of significance.

As forγindexes all the indexes in the 6 regions met the statistical test of significance indicating thatγconvergence has taken place during the period of 2000 to 2015. However,the speed ofγconvergence varied more widely among the regions compared to the income subgroups. For example,the annual speed ofγconvergence ranged from 3.14% by ECNA to 0.4% by EAP.The reason for such rapid speed displayed by ECNA may be similar to the case of the high income subgroups where a majority of countries have approached 100%target in their IWS so that ranking changes among countries can take place quite frequently. In contrast, the reason of SSA displayed one of the slowest annual speed at-0.48% in spite of the fact that its annual rate of increase in IWS was the most rapid at 1.05%. In other words, the results ofγconvergence indicate that speeds of catchup among individual countries are lagging when compared to other regions. These results suggest that some individual countries in the SSA region may require special help in overcoming their water shortage in the future.The remaining 3 regions of LAC,SA,and MENA displayed similar annual speeds ofγconvergence at 1.14%,1.11%, and 1.03% possibly reflecting the fact that their 2015 IWSs were also similar at 93.9%, 90.2% and 89.19%respectively.

5 Conclusions

The key findings from this research can be summarized as follows: First,the averaged IWS for the total group of 203 countries displayed a moderate improvement at the CAGR of 0.43% during 2000 to 2015. The annual improvement rate varied by the subgroups of countries by income level and regions. Among the income subgroups, the higher is the income level, the higher averaged IWS which in turn generates the lower annual improvement rate of IWS.For example,the low income subgroup generated the most rapid annual improvement rate of IWS at 1.21%,as opposed to 0.06%realized by the high income subgroup. Similarly,those regions with low IWS generated the high annual improvement rate,while the regions with high IWS generated the lower annual improvement rate. For example,SSA with the lowest IWS of 62.16%in 2000 generated the highest annual improvement rate of 1.05%.

Second, for the total group of 203 countries, dispersion of IWS measured in CV has declined by 24.2%whileγindex has declined by 9.5% during the period. Combining reduction of dispersion (σindexes) with increasing catch-up effect (γindexes) indicate that substantial reduction in country differences of IWS has resulted, reducing the gap among countries within the total group. The speed of reduction forσindexes was almost three times faster at-1.83%per year over-0.66%per year forγindexes.

Table 6 Normalized Sigma and Gamma indexes for six regional subgroups(2000-2015).

***Significant at 1%level,**Significant at 5%level,*Significant at 10%level

Table 6 Continued.

Third, dividing the total group into four subgroups by income levers, the high income subgroup led the statistically significant annual speed of decline forγindexes followed by the upper middle, low middle and low income subgroups in that order. In other words, the sequence of income subgroups on the speed of catchup reversed the annual improvement rate of IWS measures. Forth, as for regional subgroups, the statistically significant annual speeds ofγconvergence were verified for all 6 regions. The most rapid annual speed of 3.14%was recorded in the ECNA region which contained many high income countries with high IWS measures. On the other hand,SSA region with the lowest IWS measures realized one of the slowest annual speed ofγconvergence-0.48%.

What are some policy implications for individual countries trying to improve their safe water access? Since the speed ofγconvergence or catch-up and the speed ofσconvergence or dispersion reduction vary so widely among different income and regional subgroups, individual countries should be guided by output measures determined for the appropriate subgroup where they belong, rather than from the total group of 203 countries.These output measure should also become the minimum targets to be achieved by a particular country since those estimated output measures represent the average speed of convergence from a given subgroup of countries.Otherwise,that country may fall behind with respect to its peer countries within that particular subgroup.

On a global scale,the Sustainable Development Goal of providing safe drinking water to 100%of population will require income and regional differences discovered in this analysis will need to be further diminished. And eventually,differences existing between and within individual countries will also need to be diminished as well.To elaborate, the region of SSA is used as a case of needing urgent improvement in the future. In spite of the most rapid annual improvement rate of 1.05%for IWS,the average IWS for the 48 countries in SSA is placed at the last position at 72.68%in 2015 among the 6 regions. There are several constraining factors which are more and less uniquely associated with SSA.First,growing urban population in SSA is projected to nearly triple by 2050, increasing from 414 million to over 1.2 billion (Hopewell and Grahan, 2014). Thus, during 1990-2015,urban access to water increased by only 4% (Ndikumana and Pickbourn, 2017). Even worse, the proportion of urban dwellers with access to piped water in their home declined from 43% to 34%(Hopewell and Grahan,2014). Another constraining factor effecting more rural dwellers in Africa is the fact that agriculture accounts for 81 percent of water withdrawal, industry for 4 percent, and municipal services (largely drinking water)for the remaining 15 percent (CABRI,2017). In spite of somewhat faster rate of improvement in rural IWS,there still is a significant rural-urban gap where only 56%of the rural population has access to IWS,compared to 87%of the urban population.

In comparing averaged percentage of rural population with access to water during 1990 to 2010 population in SSA at 57.1%is substantially lower than 79.6%for other developing regions. As for the averaged percentage of urban population with access to IWS in SSA at 85.5%is again lower than 94.1%in other developing regions(Ndikumana and Pickbourn,2017). Among individual countries within SSA,urban rural ratio on access to IWS ranged from the high of 6.6 in Ethiopia to the low of 1.02 in Uganda with an average ratio of 1.8 for the total 47 countries (Ndikumana and Pickbourn, 2017). Still another constraining factor deals with the existence of significant disparities by income in SSA where 64%of the poorest quintile of urban population have access to IWS compared with 94%of the wealthiest quintile. (Hopewell and Grahan,2014).

Finally,one of the most critical constraint facing countries in SSA deals with the lack of financing to expand and maintain water infrastructure. One of the most widely quoted estimate for capital investment to achieve 2030 safety managed water supply for all as defined in SDG 16.1 by Hutton and Varughese (2016) indicates that approximately 3 times higher investment will be needed than the investment for the basic improved water service level defined in the Millennium Development Goals(MDG).For example,the annual capital investment projected is $6.9 billion for basic service and $37.6 billion for safety managed water supply each year during 2015 to 2030. When sanitation is added, the combined annual capital investment for both water and sanitation increases from$28.4billlion for basic service to$86.9 billion for safety managed service annually. The former represents about 0.1% of Combined Gross Products of 140 countries, while the latter represents 0.39%. The capital investment requirement for SSA is estimated to be about 6 times higher at 0.64%of the regional Gross Product of SSA for basic service and 2.01% for safely managed service. In comparison to developed region where the required capital investment is extremely low at 0.02%for basic service and 0.12%for safety managed service,these estimates show clearly that burden of financing water and sanitation for SSA region is much greater in the range of 17 to 32 times. In addition to capital investment, subsequent years will bring about increasing need for operating and maintenance costs. Toward 2030, it is projected that operating and maintenance cost will begin to exceed the annual capital expenditure,which will represent additional financing needs specially if efficiency of operating institutions and tariffs policies are not strengthened to cover operating and maintenance costs. (Hutton and Varughese,2016).

In view of very high financial needs, it is not surprising that 33 out of the 38 African countries surveyed report that current finance is insufficient to meet targets established for drinking water and sanitation. Among 11 African countries reporting to the survey had widely variable public expenditures on sanitation and drinking water ranging from 0.13%to 1.78%of GDP.For several countries,their household contributions are reported to exceed 40%of the total expenditure for water and sanitation,placing a heavy burden on the poor people in these countries(World Health Organization,2015).

To compensate for the financing gap,many external support agencies such as European Union Institutions,International Development Agency of the World Bank,African Development Bank and many OECD countries provide financial aid and technical expertise. For example, in 2012, aid commitments for water and sanitation for Africa comprised 7.6%of total development aid at$4.4 billion. SSA region received about 90%of this total or$4 billion in commitment.

What are some specific recommended actions for SSA?First,the level of investment in water infrastructure needs to substantially increase as a percentage of GDP.Indicative targets are more than 1%of GDP(GIZ,2019).Second,ensure there is a clear sector structure in national,regional,and local systems. Appropriate coordination and monitoring practices must be in place. Third,improve on efficiency,particularly utility performance in the urban sector and local government in the rural sector. Fourth, generating cash from tariff revenue to support capital investments made into the water sector is critically important. Tariff systems should allow for cross subsidization from high income to low-income customers for a minimum quantity of drinking water. Fifth,increase finance through innovative sources and emphasize the larger role private service providers can play.Sixth,recognize that household expenditure is likely to remain a significant portion of water sector expenditure,which means that the poor are paying more for their services and from their pockets(CABRI,2017).

This paper has several limitations, however. In addition to income and regional categorizations, there are several other factors such as urban versus rural, gender, population, age and other environmental factors that would yield some additional insights. Furthermore,by linking improving water shortage to some of these other dimensions of economic,social,and environmental issues,future research can aim to generate greater synergistic effects toward a more sustainable and balanced development within country as well as across countries.

In spite of many limitations,we hope that convergence analysis we have presented in this paper may persuade others to consider including the reduction of water shortage gap to be an equally important improvement goal in their future studies(World Bank Group,2017;Cha et al.,2016;Hutton and Chase,2017).

Acknowledgements

We thank So-Eun Kim,a research assistant at Gachon Center of Convergence Reference for her contribution in preparing this manuscript.