From a millennium to a sustainable water and sanitation development: were we there already?

As a vital element of public health and basic living standards, water and sanitation services (WSS) have been at the core of international debate for decades due to the persistent inequalities derived from the inability of implementing their universal access (Castro & Heller 2009). In fact, the progress in that direction has been significant over the last few decades, even with the significant differences between low- and middle-income and high-income countries in the access to basic WSS.

Nowadays, however, with the Sustainable Development Goals (SDGs) proposed by the United Nations (UN), by seeking to ‘Ensure availability and sustainable management of water and sanitation for all’, SDG 6 has been on the agenda of all its Member States (Howard 2021). Nonetheless, the impact of COVID-19 on the life and survival of populations worldwide has taken its toll across various levels, which resulted in a divergence from the proposed SDGs (United Nations 2020).

Still, the upgrade from the original Millennium Development Goals (MDGs) to the SDGs was a much-needed enhancement since those initial eight goals did not consider WSS as an individually critical component of worldwide sustainable development. Nonetheless, understanding if the efforts over the fifteen years of MDG pursuance were enough to bridge the gaps between high-income and low- and middle-income countries and decrease the existing asymmetries between 2001 and 2015 is fundamental to grasp the success of their proposal. Therefore, studying whether low- and middle-income countries converged towards or deviated from what we now recognise as the SDG 6 is a key indicator of the MDGs' triumph.

In this work, we follow the reasoning of Pereira & Marques (2021) and propose a similar benchmarking exercise based on a state-of-the-art technique to assess the convergence of 123 low- and middle-income UN Member States regarding their performance in terms of the desirable and undesirable targets of the MDGs concerning WSS between 2001 and 2015 (the first and last full years of the MDG agenda, respectively). These targets correspond to the WSS indicators, which are only considered by MDG 7 (‘Ensure environmental sustainability’), and concern the Proportion of total water resources used (target 7.A, indicator 7.5), the Proportion of population using an improved drinking water source (target 7.C, indicator 7.8), and the Proportion of population using an improved sanitation facility (target 7.C, indicator 7.9). Data were retrieved from the UN's official site for the MDG indicators (http://mdgs.un.org/unsd/mdg/Data.aspx). Ultimately, we understand the impact of several contextual variables in our results and compare them with the ones obtained by the aforementioned authors for the 2016–2017 SDG 6 period. To the best of the authors' knowledge, no comparable applications were found in the literature.

Focusing on the idea of convergence, economically, it is recognized as the ‘catch up’ effect of low- and middle-income countries regarding high-income countries (Romer 1994). Resting on this definition, we arrive at the two fundamental concepts of convergence: - and -convergence.

Note that all efficiency change, average technological change, average worst practice change, -convergence, and -convergence are lower than, equal to, or greater than one. In particular, an efficiency change, average technological change, or average worst practice change value greater than one denotes improvement, a -convergence value greater than one denotes convergence, and a -convergence value greater than one denotes divergence, and vice versa.

The aforementioned convergence concepts are traditionally measured by econometric methods and single productivity measures. Therefore, more in-depth results require a multi-input multi-output setting, which has been the focus of the work of Horta & Camanho (2015), for instance. Nevertheless, the desirability and undesirability of inputs and outputs is a reality in the real world. Accordingly, when conducting performance assessments, such as the one proposed here, not only such a setting is essential, but also assuming the inputs as unitary is a requirement.

In a performance assessment context resting on a multi-desirable multi-undesirable output (MDMUO) setting, the computation of the distances mentioned above is possible through the so-called ‘Benefit-of-the-Doubt’ approach (Cherchye et al. 2007) – based on the popular non-parametric frontier technique of Data Envelopment Analysis – to construct a composite indicator using directional vectors. Examples of these accepted aggregation, interpretation, and communication tools include the Environmental Performance Index (Emerson et al. 2012), the Climate Change Performance Index (Burck et al. 2012), and the Human Development Index (United Nations 2013), whereas instances of applications of the directional ‘Benefit-of-the-Doubt’ approach are abundant (see, e.g., Oliveira et al. (2019), Rogge & van Nijverseel (2019), Silva Portela et al. (2019), and Pereira et al. (2021a, 2021b)). The two Data Envelopment Analysis models used to compute the distances to the BPF and the WPF based on the three MDG 7 indicators, generate the values of the ⁠, ⁠, and ⁠, and then serve as the basis for calculating the two types of convergence have already been described by Pereira et al. (2021a, 2021b), alongside further methodological details.

As a matter of fact, Data Envelopment Analysis is the most employed non-parametric frontier technique in WSS performance assessments (Berg & Marques 2011), with plenty of distinct applications (see, e.g., Carvalho & Marques (2011), Carvalho et al. (2015), Cetrulo et al. (2020), de Witte & Marques (2010), Ferreira da Cruz et al. (2012), and Marques et al. (2014)). Nevertheless, as far as the authors are aware, there are no applications studying convergence in the WSS sector, apart from Pereira & Marques (2021), but these authors do not consider the MDGs.

First, among the eight UN MDGs, there are none concerning WSS specifically. A closer look allows the acknowledgement of the MDG 7 (‘Ensure environmental sustainability’) and three of its ten indicators as the ones that can be adequately considered for this analysis:

• Target 7.A: indicator 7.5 (Proportion of total water resources used);

• Target 7.C: indicators 7.8 (Proportion of population using an improved drinking water source) and 7.9 (Proportion of population using an improved sanitation facility).

We can now see the MDMUO setting as being comprised of one undesirable (7.5) and two desirable indicators (7.8 and 7.9).

Second, the 123 UN Member States considered in this analysis result in 46 from the Sub-Saharan Africa, 15 from the Northern Africa and Western Asia, 13 from the Central and Southern Asia, 12 from the Eastern and South-Eastern Asia, 28 from the Latin America and the Caribbean, 2 from the Oceania, and 7 from the Europe and Northern America UN regional groupings. These nations correspond to the same set as the one used by Pereira & Marques (2021) for comparison purposes. Bear in mind that the results are analysed from global and regional perspectives, but both concern a single worldwide frontier, since the reduced number of countries in some regional groupings prevents the generation of suitable regional metafrontiers.

Third, Table 1 contains the key descriptive statistics of the three indicators based on the selected UN Member States. It is immediately clear that the undesirable indicator 7.5 experienced a reduction, on average, from 2001 to 2015; on the contrary, desirable indicators 7.8 and 7.9 showed an increase, on average, in the same period.

At last, the application of the ‘Benefit-of-the-Doubt’-based performance assessment methodology for studying convergence in the selected MDMUO setting between 2001 and 2015 (the first and last full years of the MDG agenda) returned the findings presented in Table 2. Note that, following Oliveira et al. (2020), instances of missing data (retrieved from the UN's official site for the MDG indicators at http://mdgs.un.org/unsd/mdg/Data.aspx) were replaced by the corresponding values from the previous year(s); otherwise, when unavailable, we have resorted to the worst performance of the sample in that specific indicator.

Worldwide, on average, all UN Member States converged in both senses from 2001 to 2015, i.e., they were able to simultaneously decrease the performance spread surrounding the BPF (Ethiopia, Mali, Mauritania, Afghanistan, Cambodia, Lao People's Democratic Republic, and Papua New Guinea revealed an ⁠, while the Central African Republic, Comoros, Sierra Leone, South Sudan, Sudan, Yemen, Colombia, and Haiti exhibited an ⁠) and decrease the gap between the BPF and the WPF in that interval. In particular, the (Congo revealed a ⁠, while Angola, Somalia, Afghanistan, and Paraguay revealed a ⁠) and the (no countries revealed either a or a ⁠) were greater than one, denoting positive performance changes in terms of technology and worst practices; additionally, the fact that the increased at a higher rate than the is responsible for the convergence in ⁠. Regionally, the results followed the same trend, with all UN regional groupings exhibiting convergence in and ⁠. The most notable cases were the significant -convergence of the countries of the Eastern and South-Eastern Asia and the Oceania grouping, and the considerable of the countries of the Oceania grouping. Note that the Oceania grouping is only comprised of two nations, which limits the inference of further conclusions. These results are visible in Figures 1–3, bearing in mind a category classification for easier readability: changes much lower than one correspond to ‘0’, changes lower than one correspond to ‘1’, changes greater than one correspond to ‘2’, and changes much greater than one correspond to ‘3’.

From another angle, in 2001, on the one hand, the Proportion of population using an improved sanitation facility was the indicator in which more Member States were efficient and displayed the best performances; on the other hand, the Proportion of population using an improved drinking water source was the indicator in which fewer Member States were efficient, but the Proportion of total water resources used was the indicator in which the performances of the Member States were lower. In 2015, the findings were similar, apart from the Proportion of total water resources used exhibiting the greater number of efficient nations. Bottom line, from 2001 to 2015, the number of efficient low- and middle-income UN Member States improved considerably in terms of the Proportion of total water resources used, but remained relatively constant regarding the two desirable indicators, with slight performance decreases. This is in line with the results obtained by Pereira & Marques (2021) in the 2016–2017 SDG 6 period, where the access to and use of water resources revealed the best results, followed by sanitation, which also agrees with the findings of the UN (2020).

Finally, it can be said with confidence that the low- and middle-income UN Member States fully converged (i.e., converged both in and ⁠), on average, in terms of the three WSS indicators of the MDGs from 2001 to 2015. During this period, we have observed a decrease in the performance spread and the gap between the BPF and the WPF. One possible explanation to these findings can be found in the work of Cetrulo et al. (2019), where the authors address the role played by private operators in the management of WSS in low- and middle-income countries, as well as the existence of economies of scale in WSS in those countries. However, inequalities in WSS coverage in nations with low coverage due to lack of assistance were already pointed out by Cha et al. (2017), which explains the results of countries like the Central African Republic, Comoros, Sierra Leone, South Sudan, Sudan, Yemen, Colombia, and Haiti (regarding the ⁠) and Angola, Somalia, Afghanistan, and Paraguay (regarding the ⁠).

Additionally, to understand the relationship between the ⁠, the ⁠, and the and several key contextual factors, we have computed a series of bivariate correlation tests. These nine variables were selected to allow for the understanding of their impact on the three performance changes given the distinct resource and environmental conditions enveloping each nation. They included the Human Development Index and its components (Life expectancy at birth, Expected years of schooling, Mean years of schooling, and Gross national income per capita), the Proportion of total water resources used, the Proportion of population using an improved drinking water source, the Proportion of population using an improved sanitation facility, and the Population density. The statistically significant results generated by the tests returned a very high positive correlation (0.988) between the and the Population density at the 0.01 level, a low positive correlation (0.308) between the WPC and the Proportion of total water resources used at the 0.01 level, and a low positive correlation (0.369) between the WPC and the Population density at the 0.01 level. These correlations imply that best practice improvements in WSS are associated with a high population density (which coincides with locations where WSS have better quality) and worst practice improvements are associated with an increase in the use of water resources (since low- and middle-income countries tend to have low usage of water resources), and also a high population density (due to the fact that areas with higher population densities tend to have more established WSS).

In comparison with the results of the SDG 6 convergence reported by Pereira & Marques (2021), we found that from the 2001–2015 to the 2016–2017 period:

• The gap between the BPF and the WPF decreased (0.7213 vs. 0.0819), although the performance spread slightly increased (1.6406 vs. 1.2417) – this means that, although the best- and worst-performing countries moved closer from 2001–2015 to 2016–2017, there were more heterogeneous performances;

• There was a significant worst practice change improvement (1.6515 vs. 11.9095), despite the technological regression (1.1913 vs. 0.9750) – this means that, although the worst-performing countries improved substantially from 2001–2015 to 2016–2017, the best-performing countries decreased their performances on average;

• The Member States of the Eastern and South-Eastern Asia regional grouping were the only ones that, on average, increased their performance spread and the gap between the BPF and the WPF – this means that the Member States of these two regional groupings were the only ones that simultaneously displayed more heterogeneous performances and whose best- and worst-performing countries moved further away from 2001–2015 to 2016–2017;

• The Central African Republic progressed from an to an ⁠, while Mauritania experienced the opposite – this means that the former moved considerably closer to the BPF while the latter moved considerably away from the BPF from 2001–2015 to 2016–2017;

• Afghanistan continued to exhibit a ⁠, i.e., the nation continued its decline away from the BPF from 2001–2015 to 2016–2017;

• More than half of the Member States managed to significantly improve their WPC (⁠⁠), while almost 20% of them managed to significantly worsen their WPC (⁠⁠), i.e., the majority of the worst-performing low- and middle-income assessed nations were able to drastically improve their performances from 2001–2015 to 2016–2017;

• Indicator-wise, the Proportion of population using an improved drinking water source became one of the indicators in which more Member States were efficient and displayed better performances, bearing in mind the differences and similarities among the indicator sets in the two analyses.

Essentially, between MDGs and SDG, the low- and middle-income UN Member States, on average, evolved from 2001 to 2017, mainly because of the decrease in the gap between the BPF and the WPF. This was due to major performance improvements, shown by the increase in the worst practice change. Particular cases of nations in the Sub-Saharan Africa and the Central and Southern Asia regional groupings have shown significant progress or regression in their respective contexts, with Afghanistan being the only Member State that exhibited very poor technological progress. Furthermore, a higher percentage of the population used an improved drinking water source.

The convergence analysis of the UN Member States in terms of the MDGs impact on WSS revealed that, despite the lack of clear focus on the sector by the MDGs in a similar way to the SDG 6, low- and middle-income nations showed signs of being fully convergent from 2001 to 2015. In fact, on average, all UN regional groupings were able to decrease the performance spread of their Member States and decrease the gap between the BPF and the WPF.

Fundamentally, regardless of the regional grouping, all Member States were able to, on average, converge in both senses. Still, when compared to the results of the SDG 6 in the 2016–2017 period, nations belonging to the Eastern and South-Eastern Asia regional grouping deteriorated their performances, despite the noticeable improvements in the worst practice change and the gap between the BPF and WPF. Additionally, the Proportion of population using an improved drinking water source was the indicator with the highest performance growth. Bottom line, we were close, but we became closer.

In the end, there should be a higher focus on political engagement towards funding targeting WSS coverage in the areas of low- and middle-income countries with a low population density. Likewise, more than just regulatory incentives, seeking partnerships with private operators and pursuing economies of scale should be two important factors to consider when developing WSS in those countries, particularly those of the Eastern and South-Eastern Asia (e.g., Malaysia, Vietnam) regional grouping.

The main limitations of this work are related to data availability and missing data concerning the UN database. As a matter of fact, efforts from the UN on this end should be made in order to provide more key data while maintaining its accuracy and reliability. As future research, understanding the year-on-year efficiency, technological, and worst practice changes of the UN Member States regarding their WSS would be an interesting and enlightening possibility.

All relevant data are available from an online repository or repositories (http://mdgs.un.org/unsd/mdg/Data.aspx and http://hdr.undp.org/en/composite/HDI).