Interlinkages between Indicators of Sustainable Development Goals: Evidence from Seven Low Income and Lower Middle-Income Countries

In 2015, 17 Sustainable Development Goals (SDGs) were adopted by member states of United Nations as a blueprint to achieve a better and more sustainable future for all. Up to date, the global indicator of framework of SDGs has 247 indicators (including repetitive indicators), which monitor progress of achieving the SDGs. The interlinkages and integration of SDG indicators have attracted attention from scholars and practitioners, however, there is limited existing knowledge of the interlinkages and interactions between SDG indicators in low income and lower middle-income countries. This article therefore uses data from seven low income and lower middle-income countries to analyze the interlinkages between SDG indicators 9.2.1 and 9.4.1, which monitors the level of industrialization and energy efficiency in manufacturing sector respectively. It is widely believed that higher level industrialization will reduce energy emission per manufacturing value added. However, the results suggest that beyond traditional knowledge, in the seven selected countries, SDG indicators 9.2.1 and 9.4.1 are positively associated with each other. A possible reason of this unconventional finding is the country context, as in low income and lower middle-income countries the manufacturing industries are usually not energy-extensive, although not technology-intensive either. This article also calls for holistic and integrated approaches to use and interpret SDG indicators in line with the national and regional context.


Background
In the United Nations (UN) Sustainable Development Summit 2015, the 2030 Agenda for Sustainable Development (hereafter '2030 Agenda') was adopted. A core component of the 2030 Agenda is a group of 17 Sustainable Development Goals (SDGs), covering a wide range of dimensions of sustainable development such as poverty reduction, public health, food supply and environment sustainability. Since the initiation, SDGs have started to become a guidance or tool of practice and scientific research in the field of sustainable development (Biermann et al, 2017). Also, there has been an increasing academic and practical attention to the interactions between different SDGs and target and indicators under SDGs.
Interlinkages between and integration of SDG indicators have attracted attention from scholars and practitioners. For example, Waage et al. (2015) classify the 17 SDGs into three layers: 'natural environment', 'infrastructure' and 'wellbeing', in which the delivery of 'wellbeing' goals are supported by SDGs at 'infrastructure'. Although the importance of each indicator is observed, Waage et al. (2015) also note that there is only one goal is related to health in the 17 SDGs rather than three goals related to health in eight MDGs. Griggs et al. (2014, P48) find that 'Socioeconomic development and global sustainability are often used as being in conflict because of trade-offs…', which is a key reason of developing relevant integrated models or frameworks to analyze, monitor, and evaluate different SDGs and their indicators. The emerging concept 'Water-Energy-Food Nexus' (e.g. Simpson & Jewitt, 2019) can be considered as a typical example to integrate different SDG indicators and analyze the interactions between them (Biggs et al., 2015). Similarly, an environment-economic-health nexus is provided by Cerf (2019) to improve universal health coverage and assist to achieve SDG 3.
The practical implementation of SDGs also demonstrate trade-offs, prioritizations, and negotiations. For example, Wong & van der Heijden (2019) find that the avoidance of conflicts may actually hamper the implementation of SDGs, as some conflicts such as those related to long-standing issues may be necessary for achieving integration. Such avoidance of conflicts is possibly because of actors have limited knowledge of the relevant targets and actions (Wong and van der Heijden, 2019), and this is also related to SDG indicators. For example, according to IAEG-SDGs (2019), data for Tier 2 SDG indicators are not regularly produced by countries so that they may not be familiar to country actors, while Tier 3 SDG indicators even do not have internationally established methodology or standards (Note 1). Taking a small island country for case study, Adshead et al. (2019) emphasize the important role of infrastructure in delivering SDGs, as well as the trade-offs between indicators when making investment and policy intervention decisions. Similarly, Schroeder et al. (2019) find that although synergies that can be created through circular economy across SDGs and their targets, trade-offs exist between different SDGs and their targets, especially in the field of environment, decent work, and human health. Such trade-offs and prioritizations reflect and are also reflected by the dynamic and interdisciplinary nature of SDG indicators, as well as the complicated political process and international negotiations of sustainable development (Hák et al., 2016;Nilsson et al., 2016).
Using a holistic approach, Moldavska and Welo (2019) incorporate SDGs into corporate sustainability assessment in manufacturing firms. It also points out the research prospects of '…interconnections between the criteria for sustainable manufacturing and between the criteria for sustainable world' (Moldavska and Welo, 2019, p66), which demonstrates the interdependence of different SDG indicators, especially those related to manufacturing. This is also a reason for this article to analyze the interactions between two indicators of SDG 9. Furthermore, although there are some existing studies about implementation of SDGs, there is insufficient attention to the interlinkages and interdependencies among different SDGs and indicators (Stafford-Smith et al., 2017). In addition, the national and regional context of analyzing, implementing and monitoring SDGs and relevant indicators are also important (Nilsson et al., 2016). Studies on interactions of different SDGs and indicators in low income countries and lower middle-income countries are also limited due to relatively weak statistical capacity in these countries (as often measured by indicators 17.18.2 and 17.18.3). Therefore, to enrich the relevant knowledge and bridge this gap, this paper choose seven low income and lower middle-income countries to examine the interdependence and interactions of indicators 9.2.1 and 9.4.1.

Questions, Data, and Methodology
This article examines the interlinkage between SDG indicators 9.2.1 and 9.4.1 in a few selected low income and lower middle-income countries. Both indicators are monitored by United Nations Industrial Development Organization (UNIDO), which is a UN Specialized Agency to promote industrial development for poverty reduction, inclusive globalization and environmental sustainability. SDG indicator 9.2.1 is 'Manufacturing value added (MVA) as a proportion of GDP and per capita', in which MVA per capita is reported in 2015 constant US Dollar (USD). This is a popular indicator to assess the level of industrialization of a country, as it is adjusted for the size of the economy and population. The measurement of MVA in 2015 constant USD also enables international and/or cross-year comparisons. SDG indicator 9.4.1 is 'CO 2 emission per unit of value added', which calculates as ratio between CO 2 emissions from fuel combustion and the value added of manufacturing sector. The unit of value added is also 2015 constant USD which is suitable for international and over-year comparisons. This is an indicator measuring the carbon intensity and energy efficiency in manufacturing sector, which can also show the technological progress of manufacturing production. Therefore, it is an indicator which monitors multidimensions of development such as environment, economy, and technology. This indicator has important role in environment protection because CO 2 emission accounts for about 4/5 of all Greenhouse Gas emission from fuel combustion during the manufacturing processes (Note 2) This article chooses the 2017 data of SDG indicators 9.2.1 and 9.4.1 for seven low income and lower middleincome countries, as defined by the World Bank (2020). These seven countries are Algeria (North Africa), Ethiopia (Sub-Sahara Africa), Haiti (Latin America and Caribbean), Pakistan (South Asia), Tajikistan (Central Asia), Ukraine (Europe), and Yemen (Middle East). These countries are selected mainly because of their relatively low socioeconomic status, for example, they are low income or lower middle-income countries, and have experienced social and/or economic turmoil in recent years. In addition, these countries are selected to have adequate geographic representativeness, as they are in major regions of the world in where a lot of developing countries are located. The 2017 data is chosen by this article because that is the most recent data for SDG indicator 9.4.1 at UNIDO Statistics Data Portal, although data for SDG indicator 9.2.1 is available up to the year 2019. All the data are publicly available via UNIDO Statistics Data Portal (2019).
In consideration of the research question, as well as the sample size and data, the article examines the interaction and interlinkage between SDG indicators 9.2.1 and 9.4.1 by calculating the correlations between them. With the assistance of SPSS, this article reports both Pearson 'r' value and Spearman 'rho' value to explore the correlations.

Results and Discussion
The descriptive data are presented in Table 1. The correlation coefficients between SDG indicators 9.2.1 and 9.4.1 are reported in Table 2. Data are from UNIDO Statistics Data Portal (2019). The results from Tables 1 and 2 show that in these seven countries, there is a strong positive correlation between MVA per capita and CO 2 emission per unit of value added from manufacturing sector. The Pearson 'r' coefficient is 0.81, and the Spearman 'rho' value is 0.857, both demonstrate that the correlation between the two selected indicators is strong and positive. The significance values of both coefficients are lower than 0.05, which show that the correlation coefficients are statistically significant.
These results contradict to the popular belief and conventional knowledge. This is because, as introduced in the metadata of the two indicators, MVA per capita (in 2015 constant USD) reflects the level of industrialization of a country including the technology intensity, and CO 2 emission per unit of value added (in 2015 constant USD) is an (adverse) indicator of technology intensity and energy efficiency. Therefore, it is widely believed that the higher the MVA per capita is, the fewer CO 2 is emitted for each unit of value added. However, the results in Table 2 demonstrate a different direction of correlation, which suggests that if a country's MVA per capita is higher, then its CO 2 emission per unit of value added is also higher. Such unconventional findings require suitable explanations.
A possible reason of this unconventional finding is the country context. These seven countries are all low income or lower middle-income countries. Therefore, their levels of industrialization are also relatively low. This may mean that manufacturing sector in these countries may not be energy-extensive, although not technology-intensive either. To give a typical example, in countries with relatively low levels of industrialization, the dominant manufacturing industries are usually with low technology-intensity (such as production of food and beverage). Some of these low-technology intensity industries, on the other hand, do not emit large amounts of CO 2 either. In contrast, countries with higher MVA per capita, although have relatively higher levels of industrialization, may also host more energy extensive industries at meantime. A typical example in this seven countries is Ukraine. It is a country which has a strong manufacturing sector (reflected by its MVA per capita, an obvious outlier of these seven countries). However, as its manufacturing is heavily inherited from the former USSR period, the military industries have a strong presence in Ukraine (e.g., Chandler, 1996), which could be an important reason of the high CO 2 emission per unit of value added, as military related industries are often energy extensive and pay relatively weaker attention to environment protection.
The above analysis also leads to further thoughts about whether SDG indicators can exactly measure the issues as planned. This article provides evidence that SDG indicators may not be able to properly reflect the situations as planned in the metadata, especially if they were used in a separated approach without paying sufficient attention to the interlinkage and interaction with other SDG indicators. For example, SDG indicator 9.4.1 aims to measure the energy efficiency and the level eco-friendly production in manufacturing, but if related indicators such as SDG indicator 9.2.1 were not taken into consideration, solely reliance on SDG indicator 9.4.1 may lead to inaccurate interpretation such as 'energy efficiency in Ukraine's manufacturing sector is lower than in Yemen, because Table  1 shows that its CO 2 emission per unit of value added is much higher than Yemen's manufacturing sector'.
This article finds evidence to support the argument that country and regional context should be considered when monitoring and interpret SDGs and relevant indicators (e.g. Nilsson et al, 2016). Country context may vary sharply even within the same country group (e.g. lower middle-income countries). In this article, Ukraine is an outliner although it is a lower middle-income country, which reflects that the level of industrialization and general socioeconomic development in Europe (including the former USSR countries in Europe) is higher than many other parts of the world. This may suggest that the interpretation of SDG indicators in European countries could be significantly differ from low income countries in Africa, which may have much lower level of industrialization and need longer periods for industrial upgrade.

Conclusion
This article uses data from UNIDO Statistical Data Portal to explore the interlinkage and interactions between SDG indicators 9.2.1 and 9.4.1 in seven low income and lower middle-income countries. The results show that being different from popular belief and existing knowledge, in these seven countries, the two indicators are positively and significantly correlated with each other. That means, the higher level of industrialization is likely to coexist with lower level of energy efficiency. Some explanations to this unconventional result are therefore provided.
The analysis also leads to further thoughts and discussions on the plausibility and suitability of SDG indicators. This article argues that the use of SDG indicators should take the country and regional context into consideration. Also, holistic and integrated approaches are recommended to analyze and interpret data and results of SDG indicators. For example, when interpreting the data about energy efficiency in manufacturing (SDG indicator 9.4.1), it is suggested to take the level of industrialization and manufacturing productivity (SDG indicator 9.2.1) into account. This article has an objective to raise the awareness of using SDGs in holistic and approaches in line with the national and regional context.
This article, as a preliminary practice-oriented research note, is not without limitations, which leaves spaces for future research. Due to the limited sample size, more sophisticated statistical methods are not used in this article. Future studies may be able to expand the sample size and use more sophisticated statistical analysis. Moreover, although the countries selected have experienced social and economic turmoil in recent years, this article has not further discussed whether and how such turmoil may affect the industrialization and manufacturing development in these countries, and whether and how the interaction and interlinkage between SDG indicators are affected by social and economic turmoil in these countries. For example, future research may investigate how the recent domestic military conflicts in Ukraine and Yemen affect the industrialization and energy efficiency improvement in these two countries. This article has an aim to raise the awareness of the interlinkage between SDG indicators and inspire discussions on the impact of country context on SDGs and relevant indicators.
In addition, comparisons of interaction and interlinkage between SDG indicators across countries and country groups (such as among low income, lower middle-income, upper middle-income, and high income countries; or among Sub-Sahara African, Central Asia, and Latin American and Caribbean countries) could also generate interesting findings. This article lays out a foundation for the author and other researchers to conduct such comparative studies in the future. Intra-group disparities also deserve sufficient attention when exploring the interlinkage between SDG indicators.