Introduction

The United Nations adopted the sustainable development goals (SDGs) in 2015 as a standard framework to measure and monitor sustainable development through 17 goals, 196 targets, and 231 unique indicatorsFootnote 1 that are refined each year to help countries reach this objective by 2030 (UN 2015; UNSD 2020; UN 2016; Mavhura 2019). The framework carries on the momentum initiated by the millennium development goals (MDGs) that were adopted in 2000 as a guideline with 8 targets for eradicating extreme poverty by 2015. However, the SDGs provide a more comprehensive evaluation of the sustainability of human development in all environmental, social, and economic dimensions (Woodbridge 2015). Furthermore, the SDGs were devised to provide standardised metrics that quantify and decompose the abstract concepts of sustainability to measurable factors. This makes it possible to compare the level of sustainable development among different case studies and time frames. The SDGs indicators are measured based on computable data such as statistical and census data. For instance, indicator 1.5.4 measures the resilience of poor people to extreme events by calculating the number of local governments that implement local disaster risk reduction strategies in line with national disaster risk reduction action plans (UNSD 2020).

The SDGs have been universally adopted by different levels of governments, research, and industry entities as a joint framework to assess sustainability and resilience concepts. Several cities have implemented this framework in their policymaking and everyday practices. As an example, the Global Vision | Urban Action program was launched in July, 2018 by the New York City Mayor’s Office for International Affairs to apply the SDGs, specifically as an international language to communicate sustainability innovations and overcome challenges (NYC Mayor’s Ofce for International Afairs 2018). Moreover, in the Netherlands, the city of Utrecht declared itself a “Global Goals City” in which the SDGs are applied as an opportunity to achieve healthy urban living (Keranidou et al. 2018; Deininger et al. 2019). The SDGs have generally been firmly anchored in European countries as the heart of policymaking. Eurostat, the statistical office of the European Union (EU), is responsible for monitoring progress being achieved for the SDGs in the EU (EUROSTAT 2021). Referring to Asian institutions, IGES (Institute for Global Environmental Strategies) in Japan is a pioneer in studying the links between the SDGs targets for enhancing political decisions in the Asia–Pacific region and globally (Zhou and Moinuddin 2017). Despite the tremendous reforms made possible by the SDGs, there are still opportunities to improve and refine the framework.

One major trade-off relates to the inclusion of a comprehensive measurement of resilience in the SDGs framework. Resilience, which is the ability to persevere and keep functioning after an extreme event or a disaster (Siambabala et al. 2011), has been argued to be a fundamental and extremely critical prerequisite for achieving the SDGs (GRP 2020; Bhamra 2015; UNDP 2018). The SDGs framework integrates the Sendai Framework for Disaster Risk Reduction (Thacker et al. 2019) by setting some targets and indicators to evaluate the level of disaster resilience in countries. For instance, target 1.5 sets out to assess the resilience of the poor and those in vulnerable situations to climate-related extreme events and other disasters. Even one of the 17 SDGs, goal 11 (make cities inclusive, safe, resilient, and sustainable), is dedicated to benchmarking resilience. However, the framework is criticised for not being comprehensive enough to evaluate resilience. Suggestions have been made to enrich it by incorporating more resilience indicators to quantify this multidimensional concept (Diaz-Sarachaga and Jato-Espino 2019).

This is a major challenge because there is a risk that without resilience, any extreme event, shock, or disaster might make futile all efforts to achieve the SDGs. As an example, indicator 3.8.1 measures the coverage of essential health services based on a series of quantitative indicators, such as “the number of hospital beds per 10,000 people” (UNSD 2023). One major concern, however, is what happens if enough hospital beds are available, but what if the hospital was built in a flood-prone area? In this case, health services could be inaccessible and useless during times of disaster. This is an example of how overlooking resilience could affect or even reverse the SDGs’ computations and estimations of sustainability. However, it is still unclear to what extent resilience is overlooked or considered in the SDGs. Hence, this study sets out to answer the following questions:

  1. 1.

    To what extent does the SDGs framework incorporate resilience measurement?

  2. 2.

    What are the aspects of resilience that have been overlooked in the SDGs?

In a recent systematic literature review, Assarkhaniki et al. (2020) explored the literature on resilience measurement methods, frameworks, and case studies with the purpose of generating a list of baseline resilience indicators that have the potential to be integrated into the SDGs. As a result, a comprehensive baseline resilience framework (abbreviated to BRF for the purpose of this study) was developed, and it includes 318 commonly used baseline indicators extracted from the literature published between 1970 and 2020. This study compares the SDGs with the BRF to define the extent to which the SDGs integrate resilience indicators. Furthermore, it discusses the strengths and weaknesses of the SDGs in terms of measuring resilience and proposes resolutions.

This paper is structured as follows. “Resilience and sustainability” and “Joint frameworks to benchmark resilience and sustainability” sections review the literature at the intersection of resilience and sustainability. Specifically, “Resilience and sustainability“ section reviews the resilience and sustainability concepts, similarities, and differences, while “Joint frameworks to benchmark resilience and sustainability“ section explores the possibility of having a joint framework to measure the two concepts. “Methods“ section presents the research methods that help compare the SDGs and the BRF. “Results“ section presents the analysis results and then in “Discussion“ section the areas of resilience that are covered or overlooked by the SDGs are discussed. It also explores how the BRF complements the SDGs and vice versa. Finally, “Conclusion“ section concludes this research by summarising the results and proposes future related topics to be researched.

Resilience and sustainability

The growing interest in resilience and sustainability discussions has created much debate about the concepts’ similarities and differences (Xu et al. 2015; Marchese et al. 2018; Ahern 2011). Resilience is mainly referred to as the ability to prepare for, absorb, recover from, and adapt to stresses and adverse events (Ahern 2011; Siebeneck and Cova 2012; Lam et al. 2016). Sustainability, on the other hand, is defined as meeting the needs of the present without compromising the environmental, social, and economic interests of the present and future generations (The World Commission on Environment and Development 1987; Adams 2006). Sustainability, therefore, can be considered as ensuring functionality over time for many aspects of life (The World Commission on Environment and Development 1987). Based on the literature that has investigated the intersection of resilience and sustainability, five different types of relationships between the two concepts can be identified (Fig. 1).

Fig. 1
figure 1

Relationship between resilience and sustainability based on the literature

Full size image

The first strand of literature considers resilience as a component of sustainability (Marchese et al. 2018; Roostaie et al. 2019). It can be seen in the model presented by Allen et al. (2019) that it evaluates the sustainability of food systems and takes into account the food crisis and environmental challenges. They suggest deploying an assessment tool that uses indicators to evaluate the sustainable food system, including indicators to measure resilience to drivers that make the system vulnerable. This approach has also been widely addressed by multilateral organisations. Asian Development Bank (ADB) uses a definition of resilience that builds on this nexus between sustainability and resilience: resilience is the ability of countries, communities, businesses, and individual households to resist, absorb, recover from, and reorganise in response to disruptive events, without jeopardising their future socioeconomic viability (ADB 2013). The Intergovernmental Panel on Climate Change (IPCC) considers resilience as a prerequisite for sustainability, which can be enhanced through proper disaster risk adaptation (O’Brien et al. 2012). Following the heated debate among multilateral organisations on the role of resilience in achieving sustainability, resilience became a common term and part of the goal setting after 2015, such as the United Nations sustainable development goals (SDGs) (Tanner et al. 2017).

In contrast to this, the second strand of literature conceives of sustainability as a component of resilience with an emphasis on the key role of managing a system sustainably so that it becomes resilient (Chapin et al. 2009; Marchese et al. 2018). On the other hand, some research examines resilience and sustainability as concepts with separate objectives. This strand emphasises that due to the major differences in spatial and temporal scales of the two concepts (explained in the next paragraph) (Marchese et al. 2018; Roostaie et al. 2019; Collier et al. 2013) as well as the specific features of any contexts, the emphasis might be either on resilience or sustainability to build a strong community (Lew et al. 2016). The fourth strand of research stresses the common elements between the two concepts. These elements can be commonly evaluated, such as the need for reflective capacity and flexibility of the process, and robustness (Voghera and Giudice 2019). Finally, the fifth strand of literature equates sustainability with resilience, mostly by either considering the latter as an alternative for the former, or vice versa (Green et al. 2017). For instance, Worstell and Green (2017) have created a resilience/sustainability index using resilience indicators.

Generally, both sustainability and resilience seek to improve the quality of human life (Xu et al. 2015). However, resilience is mainly judged as process oriented which relates to a short space of time before, during, or after a disaster trying to maintain the processes of a system and its functionality (Rus et al. 2018; Marchese et al. 2018). Sustainability, conversely, is understood to be a long-term outcome-oriented process and especially a system’s future functioning (Costanza and Patten 1995). The differences and similarities suggest that benchmarking future development upon either resilience or sustainability might be a poor strategy.

Joint frameworks to benchmark resilience and sustainability

Several joint frameworks have built their evaluation systems based on the relationships between the two concepts illustrated in Fig. 1. However, some of these frameworks are designed for a specific field or context. For example, Milman and Short (2008) developed their water provision resilience framework, which measures both sustainability and resilience of urban water systems. Similarly, Saunders and Becker (2015) introduced a framework to measure resilience and sustainability to an earthquake in New Zealand, which cannot be used in other contexts. Other joint frameworks evaluate either resilience or sustainability despite accepting one as a component of the other. For example, in the framework devised by Jarzebski et al. (2016) for assessing the social–ecological system sustainability of community-based forest management, resilience indicators have been applied because they are fundamental to achieving sustainability. Although aiming to evaluate both concepts, the framework only incorporates resilience indicators. In a similar way, Green et al. (2017) measured eight qualities of resilient food systems as a sustainability/resilience index.

Generally, there are two main limitations of the reviewed joint frameworks. One is their context-specific nature, with a limited number of indicators to assess a specific issue (earthquake, water provision, etc.), and the other is their failure to assess both concepts. Measuring resilience to a specific type of event in a precise spatio-temporal context is the most common approach (Mihunov et al. 2019; Moghadas et al. 2019; Song et al. 2018; Green et al. 2017). To apply this approach, two main questions should be answered: (i) “resilience of what?” which limits the spatial scales (e.g. household, community, country, etc.); and (ii) “resilience to what?” which defines the stress that people are exposed to (e.g. flood, drought, outbreak of a disease, etc.) as well as the temporal scale (e.g. pre-event, post-event, and during event). This makes it possible to select a limited number of indicators making possible the measurement of resilience for the specific purpose of a case study, for instance, measuring post-flood resilience of an agricultural community. A less understood approach aims to measure baseline resilience, which provides a baseline index for resilience to any stress (from disasters to persistent stress) that we might encounter (Green et al. 2017; Tan et al. 2017; Kaye-Blake et al. 2019; Fu and Wang 2018). In baseline resilience, also referred to as general resilience, the common factors that might help a group of people survive an extreme event are measured, such as availability of shelter, hospital beds, or any type of facility or measure that might help a community to respond to any type of stress, ranging from a pandemic to a flood.

Among the current joint frameworks, the SDGs unify both resilience and sustainability metrics based on the baseline resilience approach (Zamboni 2017). However, this framework has been criticised for incorporating only a limited number of indicators to measure resilience; it is not enough for such a multidimensional concept. To address this discrepancy, this study examines how resilience is integrated into the SDGs, what aspects are overlooked, and how it can be improved.

Methods

In this study, to answer the research questions, the SDGs indicators have been compared with those of the most recent comprehensive baseline resilience framework (BRF) to establish their similarities and differences. Several text similarity methods can be applied to define the similarity of indicators varying from lexical matching methods to more complex semantic ones. However, text similarity metrics will generally fail in identifying semantic connections between text that originates from the similarity of the concepts rather than the words (Mihalcea et al. 2006). For instance, the SDGs indicator 5.4.1., “proportion of time spent on unpaid domestic and care work, by sex, age and location”, and one of the BRF indicators, “minimum wage rate per day” (Siebeneck et al. 2015) both address income level, which is linked to economic resilience. This semantic similarity, however, is difficult to define using automatic text similarity methods.

Here, the semantic similarity analysis of the two frameworks has been conducted in three steps. In the first step (coding), both groups of indicators from the BRF and the SDGs have been coded applying the thematic coding method using NVivo software. This method enables us to code phrases or sentences (here indicators) to describe aspects of the data (their connection with resilience variables, sub-domains, and dimensions) (Gibbs 2007). Therefore, each of the SDGs and BRF indicators are considered as a unique code. Next in the clustering step, the matrices of coding have been created based on the near neighbour (NN) method to establish the links between overlapping themes (shared variables) to form clusters (Guest et al. 2012) (Fig. 2). Finally, the results from the matrices have been analysed through the cosine similarity method to define to what extent the SDGs indicators cover the assessment of the resilience dimensions (Muflikhah and Baharudin 2009).

Fig. 2
figure 2

Conceptual graph for the first two steps of the method. First step: coding and connecting indicators of resilience from the BRF and the SDGs to variables. Second step: pairing the indicators from the BRF with those from the SDGs that are connected to similar variables, to develop clusters of indicators relevant to each resilience dimension

Full size image

First step: coding

For the coding exercise, a code tree was developed based on the connection of each of the BRF indicators with any of the resilience dimensions, sub-domains, and variables (Fig. 3). This hierarchy was used to develop the BRF (Assarkhaniki et al. 2020). Here, applying the same hierarchy for coding of the SDGs and BRF indicators makes it possible to determine the similarities between the two groups of codes.

Fig. 3
figure 3

Coding process and hierarchy for the BRF indicators

Full size image

The BRF attributes resilience as consisting of five key dimensions: social, economic, institutional, infrastructural, and environmental. The social dimension relates to the features of population and demographics, people’s health and knowledge, and communities’ collective characteristics. The economic dimension addresses the financial aspects of a society. The institutional dimension demonstrates the government’s ability to function well in times of adverse change or danger. The infrastructural dimension looks at the availability of infrastructure, and finally, the environmental dimension refers to human–environment interactions (Assarkhaniki et al. 2020).

In each dimension, there are several sub-domains related to different features of the dimension. For example, the social dimension addresses demographic features, which constitute one of the social sub-domains. At the lower level, there are some variables in every sub-domain that relate to different categories of features that reflect the sub-domain. For example, for the demographic sub-domain, there are some variables such as population change and population size. The variables are measured by indicators that form the lowest level of a code tree. Here, for coding the resilience indicators, this recently developed hierarchy of variables, sub-domains, and dimensions for resilience indicators has been followed (Fig. 4).

Fig. 4
figure 4

A sample coding of the SDGs indicators by employing the same hierarchy used in coding the BRF indicators

Full size image

After coding the BRF indicators (see Appendix 1 for the codebook), all 231 SDGs indicators were coded applying the same method and set of codes established in the previous stage (Fig. 3). Similar coding connects each of the SDGs indicators to the resilience variables (Fig. 4). To avoid any bias in the code definition, as mentioned before, each of the SDGs indicators (the entire sentence) was considered as a code. Later, the codes (indicators) were connected to a variable based on the general aspect of resilience that they measure. For instance, in Fig. 4, SDGs indicator 9.2.2, which measures some aspects of employment, is connected to workforce. Later, based on the similarity of the SDGs and BRF indicators for each variable, the coding matrices were devised applying the NN method. NN is a clustering technique that categorises unknown data points (in our case indicators) based on their nearest neighbour (here variables) whose class is already identified (Bhatia and Vandana 2010).

During the coding process for the SDGs indicators, where an indicator has relationships with more than one variable, more codes were defined for the indicator to reflect the SDGs’ coverage of all the relevant resilience sub-domains. For instance, the SDGs indicator 2.1.2 (“Prevalence of moderate or severe food insecurity in the population” (Siebeneck et al. 2015)) has been connected to two sub-domains of social resilience: “health” and “community strength” (see Appendix 2).

Second step: clustering

Following the completion of the coding process, by connecting the SDGs and BRF indicators to similar variables, sub-domains, and dimensions, five indicator clusters were formed. Each cluster is related to one of the resilience dimensions and incorporates two group of indicators, specifically those from the SDGs and those from the BRF (Fig. 2). Later, a matrix was developed for each cluster to find the shared and unshared variables between the two groups of indicators (Fig. 5).

Fig. 5
figure 5

Matrix coding for five clusters developed after the coding process. Numerical values in colour tiles represent the number of available indicators for each variable (recall the acronyms from Figs. 8–12 or from Appendix 1)

Full size image

The matrices are an arrangement of \(V=\left\{{v}_{1},\ldots {v}_{n}\right\}\) which functions as an array of variables in their columns (n columns), while \(F=\left\{{f}_{1},\ldots {f}_{m}\right\}\) is an array of frameworks in their rows (m rows). Consider that \({I}_{\text{No.}}\left(f, v\right)\) denotes the number of indicators that are available for \(v \in V\) in \(f \in F\). Since each row can be mathematically considered as a vector (Livesley 1964), the ith framework can be represented as an n-dimensional vector:

$${\overrightarrow{f}}_{i}=\left({I}_{\text{No.}}\left({f}_{i}, {v}_{1}\right),\ldots {I}_{No.}\left({f}_{i}, {v}_{n}\right)\right).$$

In our case, we have two frameworks (the SDGs and the BRF) and n variables related to each of the five clusters that form 2 × n matrices for each cluster (Fig. 5). As discussed earlier, each matrix also forms two vectors; let us consider \(\overrightarrow{R}\) for the vector that represents the BRF’s row and \(\overrightarrow{S}\) for the SDGs:

$$\overrightarrow{R}=\left({I}_{\text{No.}}\left(r, {v}_{1}\right),\ldots {I}_{\text{No.}}\left(r, {v}_{n}\right)\right), \quad \overrightarrow{S}=\left({I}_{\text{No.}}\left(s, {v}_{1}\right),\ldots {I}_{\text{No.}}\left(s, {v}_{n}\right)\right),$$

where r and s are, respectively, the BRF and the SDGs.

The developed vectors represent the number of indicators that the SDGs and the BRF have for each variable. However, to understand to what extent the SDGs cover resilience measurement, knowing whether they have any indicators to assess each of the resilience variables (yes or no, 1 or 0) is enough. Additionally, converting the decimal values to binary reduces the bias that may occur through the coding process. Consequently, we applied the following function to \(\overrightarrow{R}\) and \(\overrightarrow{S}\) to convert decimal vectors to binary:

$$\mathcal{G}\left({I}_{\text{No.}}.\left(f,v\right)\right)=\left\{\begin{array}{ll}1 & {I}_{\text{No.}}.\left(f,v\right) \ge 1 \\ 0 & \text{Otherwise}\end{array}\right..$$

Therefore, we have:

$$\begin{aligned} \overrightarrow{{R}_{b}}=\left(\mathcal{G}({I}_{\text{No.}}\left(r, {v}_{1}\right)),\ldots {\mathcal{G}(I}_{\text{No.}}\left(r, {v}_{n}\right))\right),\\ \overrightarrow{{S}_{b}}=\left({\mathcal{G}(I}_{\text{No.}}\left(s, {v}_{1}\right)),\ldots {\mathcal{G}(I}_{\text{No.}}\left(s, {v}_{n}\right))\right).\end{aligned}$$

Because \(\overrightarrow{{R}_{b}}\) and \({\overrightarrow{S}}_{b}\) are formed from binary data (have or do not have an indicator for any variables), they can also represent sets as written below:

$$\begin{aligned}{R}_{b}=\left\{r| \mathcal{G}\left({I}_{\text{No.}}\left(r, {v}_{i}\right)\right)\ne o\right\},\\{S}_{b}=\left\{s|\mathcal{G}\left({I}_{\text{No.}}\left(s, {v}_{i}\right)\right)\ne o\right\}.\end{aligned}$$

Both forms of sets and vectors make it possible to apply a wide range of methods to calculate the similarity between the SDGs and the BRF.

Third step: evaluating the similarity of the SDGs and the BRF

In the third step, to measure the similarity between the SDGs and the BRF, we calculate the small distance between \(\overrightarrow{R}\) and \(\overrightarrow{S}\) in each cluster (Pang-Ning et al. 2005). Several methods have been devised to measure the small distance, the most popular being the overlap coefficient (Vijaymeena and Kavitha 2016), the Jaccard similarity coefficient (Alqadah and Bhatnagar 2010), and cosine similarity (Klawonn et al. 2013; Huang 2008). Following the usability of each measurement method, each has been tested based on the research objectives to select the best matching approach (Appendix 3).

Here, cosine has been selected as the method to measure the similarity of frameworks where any relationships with 50% similarity or greater represent moderate similarity, while 75% or more represent a high level of similarity (Paul 2014).

In this study, after measuring the similarity of the frameworks, the aspects of resilience that have been addressed in the SDGs framework along with the relevant goals and indicators are documented in Appendix 2 and visually summarised in the parallel coordinate plots (PCP) (Fig. 6). Finally, the resilience dimensions covered by the SDGs were ranked by calculating the weighted standard deviation (WSD) (Fig. 7):

$$\text{WSD}=\sqrt{\frac{\sum_{i=1}^{N}{w}_{i}{\left({x}_{i}-{\overline{x} }_{w}\right)}^{2}}{\frac{{(N}{{^{\prime}}}-1)\sum_{i=1}^{N}{w}_{i}}{{N}{{^{\prime}}}}}},$$

where \({x}_{i}\) is the number of available indicators in each goal for resilience dimensions, \({w}_{i}\) stands for the frequency of \({x}_{i}\), \({\overline{x} }_{w}\) denotes the weighted mean, \(N\) represents the number of \({x}_{i}\), and \({N}^{{\prime}}\) is the number of non-zero weight. Based on WSD, the goals will be categorised in classes based on having low to very high coverage for a resilience dimension.

Fig. 6
figure 6

Parallel coordinate plots (PCPs) of five resilience dimensions. Plots represent the number of indicators for each of the 17 sustainable development goals that measure each resilience dimension (e.g. the first plot on the top left shows there are five indicators in goal 1 which addresses social resilience)

Full size image
Fig. 7
figure 7

Categorisation of the SDGs based on their coverage of resilience dimensions

Full size image

Results

As elaborated in “Methods“ section, according to the similarity of the variables that each of the SDGs and BRF indicators measure, coding matrices were developed. The matrices were created based on having the framework as rows and variables as columns to evaluate the variables that are common between the two frameworks in each cluster (Fig. 5).

Each matrix coding represents the number of indicators that each of the SDGs and the BRF have for all the resilience variables, sub-domains, and dimensions. For example, the matrix coding for social dimensions reveals that the SDGs and the BRF encompass 3 and 4 indicators, respectively, to measure communication (CS.1). CS.1 is a variable for “Community Strength” which is a sub-domain of social resilience.

As some of the indicators have been repeated in different dimensions, in a matrix coding for one resilience dimension, one or two variables might represent some features of the other dimensions. For instance, the matrix coding for the infrastructural cluster integrates some variables for the “Health Infrastructure” sub-domain (HI) that are similar to the “Health” sub-domain (H) in the social cluster. This is because the same indicator, “Hospital beds per 1000 people”, has been used to indicate the provision of health services in both social and infrastructural dimensions (see Appendix 1).

In the next step, after converting the count data to binary, cosine similarity was calculated for each cluster. As an example, Cosine similarity for the economic cluster was calculated as follows:

\(\overrightarrow{R }=[\mathrm{6,4},\mathrm{1,6},\mathrm{2,1},\mathrm{1,1},\mathrm{1,6},\mathrm{2,4},\mathrm{6,6},9]\) and \(\overrightarrow{S }=\left[\mathrm{1,0},\mathrm{0,3},\mathrm{0,0},\mathrm{1,0},\mathrm{2,4},\mathrm{0,0},\mathrm{7,2},6\right],\)

\(\overrightarrow{{R}_{b}}=[\mathrm{1,1},\mathrm{1,1},\mathrm{1,1},\mathrm{1,1},\mathrm{1,1},\mathrm{1,1},\mathrm{1,1},1]\) and \(\overrightarrow{{S}_{b}}=\left[\mathrm{1,0},\mathrm{0,1},\mathrm{0,0},\mathrm{1,0},\mathrm{1,1},\mathrm{0,0},\mathrm{1,1},1\right],\)

$$\begin{aligned}Cosine\left(\overrightarrow{{R}_{b}},\overrightarrow{{S}_{b}}\right)&=\frac{\overrightarrow{{R}_{b}}\cdot \overrightarrow{{S}_{b}}}{\left|\overrightarrow{{R}_{b}}\right|\left|\overrightarrow{{S}_{b}}\right|}=\frac{\sum_{i=1}^{N}{r}_{i}\times {s}_{i}}{\sqrt{\sum_{\mathrm{i}=1}^{N}{r}_{i}^{2}}\sqrt{\sum_{i=1}^{N}{s}_{i}^{2}}}\\ &= \frac{1\times 1+1\times 0+1\times 0+1\times 1+1\times 1+1\times 0+1\times 1+1\times 0+1\times 1+1\times 1+1\times 0+1\times 0+1\times 1+1\times 1+1\times 1}{\sqrt{{15\times 1}^{2}}\sqrt{9\times {1}^{2}+6\times {0}^{2}}} \times 100=81.65.\end{aligned}$$

The calculated similarity based on the binary data shows the percentage of variables for each resilience dimension, whereby the SDGs have at least one indicator for them (Table 1). The similarity value ≥ 75% is considered high, ≤ 25% is low, and between 50%and 75% is deemed to be of moderate similarity (Paul 2014).

Table 1 Cosine similarity between the SDGs framework and the BRF for each of the resilience dimensions (overall similarity shows the average of the values for five dimensions)
Full size table

Appendix 2 demonstrates where the similarities originate from and which goals, targets, and indicators are relevant to each of the resilience dimensions. After defining the similarity of the BRF and the SDGs, parallel coordinate plots (PCPs) were developed for each resilience dimension (Fig. 6). PCPs demonstrate which goals address each of the dimensions as well as the number of indicators the SDGs have for each of the resilience dimensions. In this way, the areas of strengths and weaknesses of the coverage are indicated.

PCPs exhibit the relationship between the SDGs (middle line) and resilience dimensions (lower line) based on the number of indicators (top row) that each of the SDGs has for each resilience dimension. For instance, social PCP shows that the greatest similarity between social resilience and the SDGs relates to goal 3 with 22 indicators, which is the highest number of indicators for this dimension. On the other hand, the least similarity relates to goals 6, 8, 10, 13, 15, 16, and 17, with only one indicator addressing social resilience. Furthermore, from the social PCP, the goals that encompass zero indicators to measure social resilience can be inferred. These goals (7, 12 and 14) have no connection with social resilience.

According to the data from PCPs by applying weighted standard deviation (WSD), the goals were ranked in terms of their coverage of each resilience dimension (Fig. 7). Based on WSD, the SDGs that incorporate 3 indicators or less for a dimension are deemed to be goals with low coverage for measuring resilience. Goals with 4–6 indicators have medium coverage, goals with 7–10 indicators have high coverage, and finally goals with more than 10 indicators have very high coverage for resilience dimensions. For instance, Fig. 7 shows that for the social dimension, goal 3 has a very high coverage, while goals 6, 8, 10, 13, and 15–17 exhibit low coverage.

Discussion

The United Nations 2030 Agenda for sustainable development breaks down the broad concept of sustainability to its measurable components, specifically 17 sustainable development goals (SDGs) measured by a set of indicators (UNSD 2020). This framework also incorporates resilience indicators (Thacker et al. 2019; Tanner et al. 2017), but it is argued as being insufficient to measure this multidimensional concept (Diaz-Sarachaga and Jato-Espino 2019; GRP 2020; Bhamra 2015; UNDP 2018). To address this problem, the present study firstly analysed the similarity between the most recent comprehensive baseline resilience framework (BRF) (Assarkhaniki et al. 2020) and the SDGs. Based on the results from cosine similarity (Table 1), the overall similarity of the BRF and the SDGs amounts to 78.28% (the SDGs have at least one indicator to measure 78.28% of the resilience variables). The high overall similarity of the BRF and the SDGs (78.28%) shows that, despite the critiques, the SDGs incorporate resilience measurement convincingly. However, the level of similarity differs from moderate to high from one dimension to the other.

The highest similarity of the SDGs and the BRF relates to social (87.71%), economic (81.65%), and environmental (80.62%) dimensions. This might be rooted in this framework’s focus on three main pillars of sustainable development: social, economic, and environmental. On the other hand, the SDGs and the BRF have the least similarity for infrastructural (73.38%) and institutional (73.03%) dimensions, which is considered moderate similarity. This is despite the importance of these dimensions being recognised in the 2030 Agenda and by academia. The SDGs acknowledge the importance of infrastructural resilience by devoting SDG 9 to addressing it. Likewise, the critical role of the institutional dimension, which relates to the rules governing the operation of the whole system, is recognised for maintaining functionality in overcoming challenges and shocks (Herrfahrdt-Pähle and Pahl-Wostl 2012).

As well as quantifying the overall similarity between the SDGs and the BRF, this first part of the analysis also serves to determine the aspects (variables) of resilience that are covered or overlooked in the SDGs. The next step to understand to what extent the SDGs incorporate resilience measurement is to determine the number of indicators that each of the seventeen SDGs have for the five resilience dimensions. The results of this analysis are presented in Fig. 6, and later used to rate the SDGs’ coverage of resilience dimensions ranging from no coverage to very high. It is based on the number of indicators that they encompass to cover each resilience dimension (Fig. 7). According to the results, all 17 sustainable development goals address at least one resilience dimension. Goals 2, 5, and 11 cover the most resilient dimensions, covering four different dimensions in each. Nevertheless, goals 7 and 14 only cover one resilience dimension. The results from the coverage ranking provide a guideline for prioritising the goals when employing the SDGs to measure resilience. For instance, according to Fig. 7, for measuring social resilience, goal numbers 2, 3, and 4 have the highest priority. Goal numbers 1, 5, 9, and 11 can be considered next, and the least priority might be given to goal numbers 6, 8, 10, 13, 15, 16, and 17.

The final part of the analysis was to determine the resilience aspects that have been covered or overlooked in the SDGs. Below, the results of the analysis are discussed for each dimension. The study proposes indicators from the BRF to be included in SDGs or post-2030 agenda as a solution for overlooked aspects (variables). The proposed indicators are documented in Appendix 1.

Social resilience

The highest similarity of resilience shown by the SDGs with the BRF relates to the social dimension (87.71%) (Table 1). Social resilience indicators have been incorporated into fourteen goals, specifically goals 1–6, 8–11, 13, and 15–17 (Fig. 7 and Appendix 2). Although there is at least one indicator in the SDGs covering many of the variables relating to social resilience, some variables are still not covered by this framework. These variables are mainly associated with the “Population and Demographic” sub-domain (P&D) as shown in Fig. 8.

Fig. 8
figure 8

The SDGs’ coverage for variables of the social dimension. Green colours show variables already covered by the SDGs and orange colours show those proposed for integration into the SDGs in future reforms. Numbers represent the code for variables of each sub-domain (E&K: 1. education level and quality, 2. public expenditure on education, 3. traditional knowledge; P&D: 1. equality, 2. household features, 3. population change, 4. population structure, 5. urbanisation and density, 6. Workforce; H: 1. expenditure on health, 2. health services, 3. health status, 4. sanitation, 5. social welfare capacity; CS: 1. communication, 2. land, 3. mobility, 4. place attachment, 5. social engagement; SP: 1. community disaster preparedness, 2. cultural heritage, 3. expenditure, 4. innovation, 5. organisation; EC: 1. food capacity, 2. household economy)

Full size image

P&D measures the features related to the population that can affect resilience, from population change to household density. Some of these features have been indirectly addressed in the SDGs. For instance, the population growth rate has been used to calculate indicator 11.3.1 (“Ratio of land consumption rate to population growth rate”). Likewise, the ratio of men to women or total population has been used in several indicators such as 1.2.1 (“Proportion of population living below the national poverty line, by sex and age”) and 1.2.2 (“Proportion of men, women and children of all ages living in poverty in all its dimensions according to national definitions”). Although the data for most of the indicators related to population (see Appendix 1) has been available and used to calculate the SDGs indicators, the direct impact of these factors on resilience has been overlooked in the SDGs.

In addition to P&D, traditional knowledge (E&K.3) which retains a community’s functionality, based on, firstly, local experience, and secondly, social welfare capacities (H.5) that address the availability of shelter for vulnerable groups, has also been overlooked (Fig. 8 and Appendix 1). Incorporating the indicators to measure these two variables and the P&D sub-domain in the SDGs framework can enhance resilience assessment’s comprehensiveness.

Infrastructural resilience

The SDGs cover several aspects of the infrastructural resilience in nine goals: 1–3, 6–8, 9, 11, and 17 (Fig. 7 and Appendix 2). However, some variables are missing from this framework, including housing features (Ho.1) (e.g. quality) and the availability of temporary housing (Ho.2), which are both very critical when a disaster occurs. Assessing the availability of workforce (IW.1) in the sector of infrastructure (e.g. transport engineers) has also been overlooked, which is critical to preparing for and coping with disasters. Other crucial aspects that have also been disregarded relate to road safety (T.4) and re-supply potential (T.3) (e.g. availability of railways), efficiency in water and gas consumption (Ef.2&Ef.3), commercial establishment (EI.1) to support economic resilience, school restoration potential (Ed.I.1) for education resilience, infrastructure (ER.3) and transportation access (ER.4) for emergency response in time of disasters, and urbanisation and density (L4). All these affect emergency response in many ways, such as the lack of access to widespread and poorly planned urban areas. Public establishments (EW&S.3), for instance, open and green spaces that can be used as shelters when disaster strikes is another overlooked aspect (Fig. 9 and Appendix 1). These components are imperative to maintaining the functionality of a system as a whole and are strongly suggested to be addressed in future refinements to the SDGs or the post-2030 agenda.

Fig. 9
figure 9

The SDGs’ coverage for variables of infrastructural dimension. Green colours show variables that are already covered by the SDGs and orange colours show those proposed for integration into the SDGs in future reforms. Numbers represent the code for variables of each sub-domain (EW&S: 1. electricity, 2. plumbing and sewage, 3. public establishment; Ho: 1. housing features, 2. temporary housing availability; HI: 1. health services; L: 1. agriculture, 2. informal settlements, 3. social engagement, 4. urbanisation and density; EI: 1. commercial establishment; Ed.I: 1. school restoration potential; ER:1. emergency stations and shelters, 2. emergency workforce capacity, 3. infrastructure, 4. transportation access; T: 1. evacuation potential, 2. accessibility, 3. industrial re-supply potential, 4. road safety; C: 1. high-speed Internet, 2. telecommunication; IW: 1. public utility sector; Ef: 1. electricity, 2. gas, 3. water)

Full size image

Institutional resilience

The indicators to measure institutional resilience are found in seven SDGs: 1–2, 5, 11, 13, and 16–17 (Fig. 7 and Appendix 2). However, the SDGs do not cover the assessment of the “Disaster Experience” sub-domain (DE), such as disaster aid experience and knowledge, which are vital when dealing with upcoming challenges. Performance (G.6) of local governments and their role in nation-wide decisions, national financial (G.3) support to recover from shocks, and the role of government in developing resilience (G.1), psychological support (RR.3)—essential for victims to bounce back from shocks—and accessibility (DP.1) of government services (e.g. municipalities) are critical factors for resilience that have been overlooked in the SDGs (Fig. 10 and Appendix 1).

Fig. 10
figure 10

The SDGs’ coverage for variables of institutional dimension. Green colours show variables that are already covered by the SDGs and orange colours are those proposed for integration into the SDGs in future reforms. Numbers represent the code for variables of each sub-domain (DP: 1. accessibility, 2. mitigation; DE: 1. disaster aid experience, 2. disaster knowledge; G: 1. development, 2. environmental, 3. financial, 4. income and expenditure, 5. infrastructure, 6. performance regime, 7. social, 8. urban improvement; RR: 1. ecosystem support, 2. financial support, 3. psychosocial support)

Full size image

Economic resilience

For the economic dimension, there is 81.65% similarity between the SDGs and the BRF (Table 1). Overall, the SDGs have addressed this dimension in goals 1, 2, 5, 7–10, 12, and 16 (Fig. 7 and Appendix 2). Economic diversity (BA.3), commercial activities (BA.2), and availability of insurance (FS.1) are among the features of a resilient economy, but they have not been incorporated into the SDGs. Similarly, the costs and revenue generated by urbanisation and density (C&R.4) and energy/water (C&R.1) production and consumption, as factors exerting a negative impact on economic resilience, have been overlooked within the SDGs (Fig. 11 and Appendix 1).

Fig. 11
figure 11

The SDGs’ coverage for variables of economic dimension. Green colours show variables that are already covered by the SDGs and orange colours are those proposed for integration into the SDGs in future reforms. Numbers represent the code for variables of each sub-domain (HC: 1. household assets; E: 1. equality, 2. workforce; I: 1. equality, 2. income features; BA: 1. business features, 2. commercial activities, 3. economy diversity, 4. income and expenditure; C&R: 1. energy/water, 2. land, 3. renewable energy, 4. urbanisation and density; FS: 1. insurance; P:1. production type and features)

Full size image

Environmental resilience

Finally, the SDGs have 80.62% similarity with the BRF concerning the environmental dimension (Table 1). This dimension has been addressed in eleven goals, namely, 2–3, 5–8, and 11–15 (Fig. 7 and Appendix 2). However, some aspects related to agricultural resilience, including connectivity (A.2) which measures farm access to the internet, self-organisation (A.6), and crop diversity (A.3) are among the crucial factors required for resilient agriculture that are not assessed in the SDGs. Frequency of natural disasters (EF.4), availability of land in safe zones (EF.5), environmental violations (SI.1), and population density (SI.3) are other factors that threaten environmental resilience and have not been measured in the SDGs (Fig. 12 and Appendix 1).

Fig. 12
figure 12

The SDGs’ coverage for variables of environmental dimension. Green colours show variables that are already covered by the SDGs and orange colours are those proposed for integration into the SDGs in future reforms. Numbers represent the code for variables of each sub-domain (A: 1. assets, 2. connectivity, 3. diversity, 4. income, 5. local food suppliers, 6. self-organisation; EF: 1. air, 2. biodiversity, 3. land and sea, 4. natural disasters, 5. safe zones, 6. soil, 7. water; EP: 1. efficiency, 2. emission, 3. recycling, 4. sewage and waste; SI: 1. crime, 2. gross production, 3. population)

Full size image

As well as the indicators for the overlooked, but high priority variables that should be incorporated in future versions of the SDGs, this study proposes more indicators to enhance the comprehensiveness of resilience measurement. The indicators can also be used as an alternative for the current SDGs indicators to help cope with the data availability issue. These suggested additional indicators, as well as the most relevant goals, are documented in Appendix 1.

“Introduction“ section demonstrated the need for a more inclusive measurement of resilience through the example of SDGs indicator 3.8.1—coverage of essential health services. It had been argued that only considering the availability of health services is not enough to ensure compliance with this criterion of sustainable development. If we do not assess the resilience level of the health services provided, it is unclear whether they will be available in emergency-type scenarios. Based on the results, a more comprehensive integration of resilience indicators into the SDGs would effectively address this gap. For example, the BRF indicators such as “% Building infrastructure not in high hazard zones” or “% Commercial establishments outside of high hazard zones to total commercial establishment” could be valuable additions to the SDGs to evaluate the availability of health services during emergencies.

The results also suggest that it is not only the BRF that can be used to improve the SDGs. Matrices in Fig. 5 show the areas in which either the BRF or the SDGs have more indicators. For some of the resilience sub-domains shown in Table 2, indicators from the SDGs can be incorporated into the BRF. For the social resilience dimension, “public expenditure on education”, “communication”, “expenditure on health”, and “health status” are the areas in which the SDGs have more indicators. Similarly, for the economic, infrastructural, institutional, and environmental resilience dimensions, “equality”, “high-speed internet”, “social governance”, and “biodiversity” are, respectively, the areas where the SDGs have more indicators than the BRF. Generally, in addition to these specific areas, all other SDGs indicators that measure resilience could be added to the BRF (list of the 119 SDGs indicators to measure resilience can be found in Appendix 2 in the supplementary material).

Table 2 Resilience aspects (variables) for which the SDGs framework has more indicators than the BRF
Full size table

Conclusion

The SDGs have been criticised for, firstly, their inability to integrate resilience and, secondly, their failure to incorporate the assessment of this concept, which is a prerequisite to achieve sustainable development. Results of this study show that resilience indicators have been well covered in the SDGs. The SDGs’ high similarity (78.72%) with the comprehensive baseline resilience framework (BRF) (see Appendix 1) means that the framework has at least one indicator to measure most of the resilience aspects (more precisely, 78.72% of the variables). Despite the high level of similarity, some aspects of resilience have still been overlooked in the SDGs; they can be integrated to enhance the comprehensiveness of the framework. These aspects were determined in this study (Figs. 8, 9, 10, 11, 12) and the resolutions proposed (Appendix 1). Among the five resilience dimensions, the least similarity with the SDGs relates to infrastructural and institutional resilience. Therefore, incorporating indicators from the BRF to measure the overlooked variables of these dimensions is of the highest priority for future improvements to the SDGs. However, to enhance the comprehensiveness of the resilience measurement within the SDGs, this study suggests integrating the indicators to cover the measurement of the overlooked variables of all the resilience dimensions.

This study proposes a comprehensive list of indicators either to cover the areas for which the SDGs do not have any indicators, or to enhance the comprehensiveness of the resilience measurement for the areas with fewer indicators (see Appendix 1). The comprehensive list of indicators can also function as an alternative to the current SDGs indicators in case of data unavailability. Although many indicators from the BRF can be introduced to the SDGs, the results do suggest a two-way relationship between the SDGs and the BRF, where the former can help to improve the latter (Table 2). The five matrices in Fig. 5 illustrate the areas in which the SDGs have more indicators to measure resilience in comparison to the BRF. Appendix 2 summarises all the goals, targets, and indicators that address each of the resilience dimensions. These indicators can also be introduced to the BRF.

Demonstrating the strengths and weaknesses of the resilience measurement within the SDGs, this study contributes to advancing the purpose and relevance of the SDGs, and generally the momentum that has since 2000 accompanied the MDGs. Also, the results contribute to expediting a seamless application of the current SDGs framework in measuring resilience by ranking the goals based on their coverage of each resilience dimension (Fig. 7). It should be noted that not all goals have the same importance and contribute to measuring a specific resilience dimension. Nonetheless, some aspects of resilience might be overlooked when considering goals in isolation or only selecting a group of them. This is because the indicators to measure the resilience dimensions are scattered among several goals. Provided in this study is a guideline for authorities who apply the SDGs as their benchmark to prioritise the goals based on their needs using the ranking presented in Fig. 7.

Finally, this study also contributes to the discussions at the intersection of resilience and sustainability measurement and the efforts to develop a joint framework to measure both concepts. The findings of this research provide compelling evidence that, in light of the suggested improvements, the SDGs have the potential to serve as a joint framework to measure both resilience and sustainability.