You are on page 1of 12

Global Policy Volume 11 . Issue 1 .

February 2020
56

Measuring Sustainable Development Goals


(SDGs): An Inclusive Approach
Arman Bidarbakhtnia
Research Article

United Nations ESCAP

Abstract
Four years on from the adoption of the Sustainable Development Goals (SDGs), countries around the world seek methods that
provide simple but comprehensive narratives on their progress. Given the complexity of the agenda and its indicator frame-
work, key questions that arise are what is an appropriate method to summarize information? and how to get everyone counted?
This paper provides an overview of various existing methodologies for assessing progress towards the SDGs. It provides guid-
ance on how to choose an appropriate method that is fit for the purpose of analysis. The paper also recommends an
approach for measuring SDG progress that accounts for progress among the furthest left behind groups in order to fulfil the
leave no-one behind ambition of the 2030 agenda. Applying the new approach to data from countries in Asia-Pacific region
shows that, when disaggregated statistics are available, the inclusive measurement can significantly change the progress nar-
rative and our understanding of priorities for SDGs implementation.

Policy Implications
• Given the diversity of practices in measuring progress towards sustainable development and the complexity of the indica-
tor framework provided by the SDGs, policymakers need to make a careful choice among existing approaches. An appro-
priate approach is one that clearly answers their policy questions.
• The 2030 agenda is a people-centred development framework and its measurement should accommodate the inclusion of
disaggregated statistics from the furthest behind population groups. SDGs progress assessment should examine if all sub-
population groups are making the required progress to achieve the 2030 targets.
• Setting clear and specific national targets are an essential part of measuring progress towards sustainable development.
To make progress assessment possible, policymakers should engage with the statistical community to define specific and
numeric target values that are aspirational but within reach if existing resources are used efficiently.
• The SDGs framework is multidimensional and any progress assessment needs to provide insight on all different dimen-
sions to be useful for national planning and prioritization. Measurement frameworks should ensure that information is
aggregated and communicated at various levels in order to capture all dimensions of the SDGs.

(UNDP, 2018) was an unrivalled measure of human develop-


Measuring complex development frameworks
ment. It encapsulated Sen’s ‘capabilities’ approach (Sen,
It is staggering to think about 232 indicators1 and grasp 1985) in one measure that combined three dimensions of
where your country or region stands against 17 goals and 169 development, which aimed to understand human well-being
targets. To make it even more difficult, imagine you only have rather than economic prosperity. The index created
a few minutes to brief high-level officials of your country on immense scientific and political dialogue (Stanton, 2007)
the progress your country has made towards the Sustainable and facilitated the mobilization of resources and commit-
Development Goals (SDGs)2 and highlight a few priority areas ments for development in the national and international
for urgent action. National Statistical Systems (NSSs) around arena. Nevertheless, two important dimensions of develop-
the world have committed themselves to take-up this chal- ment (environmental and institutional) were absent from
lenge by a) producing high quality disaggregated statistics on the framework. The first time official statisticians were chal-
the SDGs indicators, and b) measuring progress in achieving lenged to measure progress for a normative and universal
the 2030 agenda for sustainable development. This paper development agenda, with quantitative and time-bound tar-
deals with the second mission and provides an overview of gets, was at the dawn of the 21st century. The MDGs put
the existing approaches and proposes an inclusive approach forth a global agreement for development with 60 indica-
for measuring SDGs progress ‘for all’. tors. Broadly speaking, there were two types of measures
used to assess progress towards the MDGs; trend analysis
(Degol and Raquel, 2011; Fukuda-Parr and Greenstein, 2010;
MDGs era
Osorio, 2008; Tabatabai, 2007; Vandermoortele, 2007), and
For one decade, prior to endorsement of the Millennium meeting the targets. The basis of assessment in the trend
Development Goals (MDGs), the Human Development Index analysis approach was the direction and pace of a country’s

© 2019 University of Durham and John Wiley & Sons, Ltd. Global Policy (2020) 11:1 doi: 10.1111/1758-5899.12774
Measuring the SDGs: An Inclusive Approach
57

progress on each of the indicators. This approach was taken in order to achieve the 2030 agenda. OECD developed a
by some countries (such as the Philippines) to develop what similar methodology in 2017 and applied it to selected
they called ‘MDGs watch’. The meeting the targets approach member countries (OECD, 2017).
was taken by United Nations Economic and Social Commis- SDGs progress measurement approaches are diverse in
sion for Asia and the Pacific (UNESCAP), Asian Development their underlying assumptions and statistical features. They
Bank (ADB) and United Nations Development Programme do not measure the same thing and should not be inter-
(UNDP) to measure progress of countries in Asia-Pacific preted in the same way. Nonetheless, this is not obvious
region based on MDG targets (when available). A series of to the general user as all methods are often referred to
tripartite reports (UNESCAP/UNDP/ADB, 2003, 2013) catego- as ‘SDG progress’. This paper aims to answer two ques-
rized countries and regions based on whether they were tions: what is an appropriate method for assessing SDG pro-
early achievers, on track, slow, or without progress based on gress? and how to ‘leave no-one behind’ in measuring
their estimated year of achieving the targets. When there progress towards the SDGs? It examines features of the
was no specific target, this approach estimated turning three major approaches taken by SDSN, UNESCAP and
point in trend to measure progress. OECD and proposes a set of criteria to be considered
when selecting a desirable method. In attempt to answer
the second question, this paper provides a detailed
SDGs era
account of the UNESCAP methodology and introduces a
The MDGs addressed essential development priorities but new factor to the index that takes into account progress
lacked some of the most fundamental dimensions. In partic- among the most vulnerable groups (when data is avail-
ular, environmental sustainability, equality, social inclusion able). Applying the new approach on data from countries
and governance were the most important goals missing in Asia-Pacific region, significant changes in progress
from the MDGs. The SDGs aimed to continue the momen- assessment results are observed.
tum generated by the MDGs and complete the unfinished The remainder of the paper is organized as follows. The
work from the previous global agreement. They also aimed following section provides an overview of the UNESCAP
to better capture the complexity of inclusive and sustainable approach for assessing progress towards the SDGs. The third
development for all. Monitoring progress towards such an sectionpresents a comparison of three major approaches.
ambitious agenda requires a complex indicator framework The fourth section introduces a modification to the UNES-
and new measurement approaches. Shortly after the adop- CAP approach to include disaggregated statistics on vulnera-
tion of the SDGs, international agencies and think-tanks ble groups. The fifth section discusses the results and
began to develop methodologies for summarizing its com- overall progress of the Asia-Pacific region towards SDGs
plex indicator framework into fewer numbers. The first two based on the new approach, and the final section concludes
attempts were the SDG index (Kroll, 2015) and SDG score- the discussion.
cards (Nicolai et al., 2015). The SDG index was based on 34
indicators (two indicators per goal) and constructed an
index that was used to compare and rank member countries SDG progress assessment: UNESCAP approach
of the Organization for Economic Co-operation and Devel-
Two indices are used by UNESCAP to assess progress
opment (OECD). The SDG scorecard was based on 2030 pro-
towards the SDGs: current status and anticipated progress
jections of 17 targets (one target per goal) and their
gap. The current status index measures progress towards
distance to the target values set by the SDGs. In 2016, Ber-
specified targets since 2000. The anticipated progress gap
telsmann Stiftung and Sustainable Development Solutions
measures the gap between the predicted value of the indi-
Network (SDSN) applied an enhanced version of the SDG
cator and specified target for 2030. Both indices are con-
index and produced the SDG index and dashboard report
structed at indicator level and can be aggregated at target
using 63 indicators from official and unofficial sources (Sachs
and goal levels as desirable. The indices answer two differ-
et al., 2016) and compared 149 countries. Since 2017, Ber-
ent questions:
telsmann Stiftung and SDSN refined the methodology for
the SDG index and dashboard report, adjusting the index • Current status: how much progress has been made since
for spillover effect which measures cross-border effects, to 2000?
take into account international responsibilities of each coun- • Anticipated progress gap: how likely will the targets be
try (Sachs et al., 2017). In 2018, the report included an SDG achieved by 2030?
trend dashboard that used time series data and enabled
countries to view progress as well as the current status.
Current status index
India applied the same methodology in 2018 to develop the
SDG India Index (NITI Aayog, 2018). Denoting indicator values for 2000 and the current year by
In 2016, United Nations ESCAP published a SDG progress I00 and Icv , and the target value for 2030 by TV, and setting
assessment report3 (UNESCAP, 2017, 2018) applying a mea- normalized values of the indicator in 2000 and 2030 at 0
sure of meeting the targets (Bidarbakht-Nia, 2017a, 2017b). and 10 respectively, normalized value for the indicator at
The report established a baseline and assessed the gaps the current year on the scale of 0 to 10 can be calculated as
which needed to be closed by the region and its subregions follows when desirable direction is clear:4

Global Policy (2020) 11:1 © 2019 University of Durham and John Wiley & Sons, Ltd.
Arman Bidarbakhtnia
58

Icv I00
Pcs ¼ jTV D
I00 j
importance attached to the indicator values should be pro-
ð1Þ portional to how recent the data is.6

10 increasing values are desirable
in which D ¼ Suppose that n data points are available on indicator I for
10 decreasing values are desirable:
a given country/region over a period of T years, and we are
For parity indicators the value is: interested in estimating the indicator value for the year t.
T ¼ tn t1 where tn and t1 are the earliest and the latest
10 jjTV Icv j
(
TV I00 j  10 if jTV Icv j  jTV I00 j years, respectively, for which data on indicator I are avail-
Pcs ¼ jIcv I00 j ð2Þ able. The time-related weights work as multipliers that
jTV I00 j  ð 10Þ Otherwise
inflate/deflate the rate of change in each period, in propor-
The above metrics are computed for each indicator and tion to the temporal distance to the target year (t). Time-re-
averaged under targets and goals to measure ‘average pro- lated weight for the ith data point for a given country/
gress made’ in achieving each of the targets and goals. region for estimating indicator values of the year t is:
Given the uneven distribution of indicators, they are
weighted proportional to the number of indicators under jt t1 j
wi ¼ ðt1 \ti \tn Þ: ð4Þ
each target when aggregating at goal level. This gives equal jt ti j
importance/weight to each target.
If the region (or subregion) has progressed since 2000,
the average of normalized values under each goal (Pcs ) pro- Setting regional target values
vides an index that is between 0 and 10. But if the region
has regressed, the value is negative and indicates the size Of the 169 SDG targets, only 30 per cent have specific (im-
of regression.5 plicit or explicit) target values. For the remainder, estimating
progress indices requires setting target values. For the
majority of such indicators, UNESCAP has applied a ‘cham-
Anticipated progress gap pion area’ approach (Bidarbakht-Nia, 2017a). The idea is to
Denoting the predicted value of indicator I for the target identify the region’s outstanding countries (top performers)
year by I30 , and value in the base year by I15 ; one can and set their average rate of change as the region’s target
approximate the progress gap by Ppg as follows when desir- rate. In other words, if the region as a whole can perform as
able direction is clear from the target: well as its champion area over 15 years (starting from 2015),
we should expect to achieve the target value by 2030.7
jTV I30 j The main challenge with the champion area approach
Ppg ¼  10: ð3Þ arises when dealing with two types of indicators:
jTV I15 j
• indicators for which there is insufficient data to estimate
In the case of parity indicators, we consider no regression
the rate of change at the country level; and
has occurred if jTV I30 j  jTV I15 j.
• indicators for which most of the countries started from a
The progress gap index only needs to be calculated for
very low level and made such rapid progress over the
indicators for which the predicted value has not reached
15 years prior to the SDGs. For these indicators observed
the target value (not expected to achieve the target). It is
growth rate cannot reasonably be applied to the future.
automatically set to 0 for other indicators. If progress or no
change is expected, the value of Ppg ranges from 0 to 10; if For these two types of indicators, an alternative approach
the predicted regression from the current level Ppg is greater is taken. Rather than using the rate of change, the top five
than 10. Ppg may be interpreted as the extra effort or accel- performers are identified based on the latest available data.
eration needed to meet the target when the value is less
than or equal to 10, and 10 – Ppg is the size of regression
when it is greater than 10.
What is an appropriate measure of progress?
To avoid the influence of outliers, an acceptance thresh- Ensuring comprehensiveness, accuracy and simplicity at the
old of minimum 2 per cent change is considered for pro- same time does not come without a compromise. Assessing
gress/regression in the calculation of both indices. In other progress towards the SDGs using the exhaustive list of 232
words, the change is acceptable only if overall change over indicators requires making many assumptions (that do not
the period is greater than a 2 per cent increase or decrease always hold) and decisions, as well as striking a balance
(depending on the actual and desired direction of change). between the simplicity of messages and accuracy of the
results. In the end, there is no single perfect method for
measuring the SDGs progress. This section highlights that
Extrapolation and imputation methods
the three major approaches that are currently employed to
Producing the two measures of progress requires prediction measure SDGs progress serve completely different purposes
of future values as well as imputation of missing values in and answer very different sets of questions. Lack of trans-
the current and previous years. These values were estimated parent methodology and metadata may have created the
using a weighted regression model that uses time-related impression that different international organizations propose
weights (Bidarbakht-Nia, 2017b), assuming that the different answers to the same question of ‘how much

© 2019 University of Durham and John Wiley & Sons, Ltd. Global Policy (2020) 11:1
Measuring the SDGs: An Inclusive Approach
59

progress have we made?’, or ‘will we achieve the 2030 tar- (country B), or significantly reduced poverty (country C).
gets?’. Nevertheless, a closer assessment of various methods UNESCAP method provides a very different ranking com-
shows that they are very different in nature and even if pared to the other two methods because it considers the
exactly the same set of data and indicators are used (which country’s progress over time. The most radical difference is
is not the case), we should not expect identical progress seen in the country B where poverty has almost doubled
narratives from them: since 2000 but is still relatively low compared to other coun-
tries of the region.
• The SDSN’s SDG index provides a starting point for coun-
The follow-up and review mechanism of the 2030 agenda
tries to compare themselves with the rest of the world.
encourages ‘regular and inclusive reviews of progress at the
For each indicator, the index measures the distance of
national and sub-national levels, which are country-led and
each country to the worst performer, as a share of the
country-driven’. So, which factors countries need to consider
total distance of the worst performer to the 2030 targets.
when investigating measures of progress? The most impor-
The index does not use time series data at the country
tant criteria to keep in mind are target values, standardiza-
level. The upper thresholds (target values) for indicators
tion and benchmarking, and aggregation.
are primarily based on expert consultations or interna-
tional commitments (when not directly specified in SDG
• Specific target values are an integral part of progress
targets), or otherwise average over top 10 per cent of the
assessment. They ensure that progress reports provide
best performers. The index may be used more as a rank-
clear measures of achievements that facilitate objective
ing tool than progress assessment measure. The progress,
policy evaluation over time. Target values have to be
however, is assessed by a measure of annual growth rate
numeric to facilitate application of scientific measures.
for each indicator as compared with the desirable growth
They must be based on realistic, yet aspirational, objec-
rate to understand whether the current speed is fast
tives that are feasible to achieve if all existing resources
enough to reach the 2030 target.
are used efficiently. Purely theoretical or politically negoti-
• The UNESCAP approach (as explained in section ‘SDG pro-
ated targets offer approaches at opposite ends of a spec-
gress assessment: UNESCAP approach’) utilizes two
trum that should be avoided, unless there is no
indices that measure: (1) progress made on each indicator
observation to rely on. Ideally, targets have to be set at
since 2000 as a proportion of total progress needed by
indicator level. The champion area provides one example
the region to meet the 2030 target; and (2) expected dis-
of setting targets based on successful (but not extraordi-
tance from targets in 2030 as a proportion of total dis-
nary) experiences from peer countries or regions.
tance the region needs to travel between 2015 and 2030.
• Standardization and benchmarking are two operations that
• The OECD approach measures the distance of each indi-
are normally conducted in one step and serve two differ-
cator from the 2030 target (applying a similar approach
ent purposes: standardization maintains comparability
to SDSN) for each country and presents it as a ratio of
and facilitates the aggregation of indicators with different
the spread of values across peer countries (using stan-
units and types, while benchmarking provides a reference
dard deviation to normalize the distance value). It results
point that facilitates monitoring of progress over time or
in exactly the same ranking as the SDSN method, as both
across countries/regions. The formula column in Table 1
methods provide the same scores when normalized on a
provides three examples of indices that use completely
scale of 0–10 (Table 2). The approach is static, meaning
different standardization methods and benchmarks. It is
that it does not account for the rate of progress by each
critical to understand that this process impacts interpreta-
country. This feature is common in both SDSN and OECD
tion of the progress. Therefore, choice of method has to
methods and differentiates them from UNESCAP
be made based on the purpose of analysis (questions
approach. The OECD approach, however, identifies the
progress assessments aim to answer).
direction of change by estimating the correlation coeffi-
• Aggregation normally takes place at two different levels:
cient between time and indicator values. A positive corre-
indicators and countries/regions. It is important for users
lation shows that a country/region is progressing in the
to investigate how indicators are weighted in each pro-
right direction to meet the target and a negative correla-
gress assessment method. The 2030 agenda gives equal
tion means that it needs a course correction to meet the
importance to all dimensions of development (17 Goals).
2030 targets.
However, targets under each goal and indicators under
Table 1 summarizes the features of the three methods. each target are not evenly distributed. At each level of
The table clearly shows that choice of progress assessment aggregation, equal importance to one level means
method depends heavily on the purpose of analysis (what unequal weight at another level (section ‘Counting every-
do we want to measure?). If we apply the three methods on one: inclusive measure of SDG progress’ provides a few
the same data, we should not expect the same outcome. examples). The progress assessment methodology has to
Table 2 shows results of applying the three methods on accommodate for such a weighting system, or hierarchi-
data for population living in poverty at less than $1.90 a cal aggregation that facilitates robust presentation of
day for three countries in Asia and the Pacific region. It results at different levels. Aggregation of indicator values
shows three different scenarios in which countries have not over countries or sub-national areas can have a major
made any progress (country A), regressed since 2000 impact on results as well as interpretations. Different

Global Policy (2020) 11:1 © 2019 University of Durham and John Wiley & Sons, Ltd.
Arman Bidarbakhtnia
60

Table 1. A comparison of three major methods used for SDG progress assessment
What is being mea- Setting target Basis of com- Static/
Method sured? Formulaa value parison dynamicb Data intensity aggregation
Icv Imin
SDSN Compared to the TV Imin Direct from Worst Static At least one Over countries: --
worst performer, how target/ performing data point
close the country is scientific country
to the target value/ top five
values
Speed required for a Average of Dynamic At least two Over indicators:
country to achieve annual data points Index- arithmetic
targets, relative to percentage average
the current speed changes Dashboard-
arithmetic
average of two
worst indicators
Icv I00
UN-ESCAP Progress a country/ TV I00 Direct from Past progress Dynamic At least two Over countries:
region has made target/top five in the same data points Median value
towards targets since rates of country/ over countries
2000, as a share of change region (unweighted)
total progress it
needs to make
TV I30
Expected distance TV I15 Dynamic At least three Over indicators:
from targets in 2030, data points Weighted
judging from a average of all
country/region’s past data series
progress (equal weights
to targets)
Icv TV
OECD Compared to the SDðIcv Þ Direct from Distribution of Static At least one Over countries:
worst performer, how target/ values across data point population-
close the country is scientific peer weighted
to the target value/ top five countries average
values
Whether or not a qðIt ; t Þ Dynamic At least two Over indicators:
country is moving in data points arithmetic
a right direction average of all
towards the target indicators

Notes: I00, I15, Icv, I30: Indicator values in 2000, 2015, current year, and prediction for 2030.
Imin : minimum value of indicator across all countries, TV: target value for indicator, qðIt ; tÞ: Spearman correlation between indictor values and
time, SDðIcv Þ: standard deviation of the indicator values.
a
Formulas are simplified for easy understanding. Readers may refer to original sources (for OECD and SDSN) and section 2 (for UNESCAP) for
more detailed explanations; bDynamic metrics measure progress over time for one specific country/region. Static metrics only measure current
status or current distance to target without indicating how much progress has been made or will be made over time.

exclude smaller ones, unless it is part of the objectives of


Table 2. Progress scores for population below international
poverty line in three countries using three different methods
the progress assessment.

2000 (%) 2015 (%) SDSN UNESCAP OECD


Counting everyone: inclusive measure of SDG
Country A 13 13 8 0 8 progress
Country B 9 16 7 9 7
Country C 46 23 6 6 6 The 2030 agenda is the people’s agenda. It puts people first
and its core principle is leaving no-one behind. The world
Notes: SDSN and OECD methods give exactly same values when cannot uphold to this principle unless the SDG progress
normalized on the scale of 0 and 10.
assessment frameworks embrace it. In other words, no pro-
gress shall be counted, unless all progress together.
measures of distribution and different weighting strate- Nonetheless, one thing that all aforementioned methods
gies may be applied. There is no perfect aggregation have in common is that they do not accommodate for the
method. One important factor to keep in mind is that the ‘leave no-one behind’ principle of the SDGs. The global indi-
2030 agenda is concerned with inclusiveness. Information cator framework for the SDGs suggests that indicators
loss occurring in the aggregation process should not put should be disaggregated, when relevant, by income, sex,
more responsibilities on bigger countries/economies or age, race, ethnicity, migratory status, disability and

© 2019 University of Durham and John Wiley & Sons, Ltd. Global Policy (2020) 11:1
Measuring the SDGs: An Inclusive Approach
61

Figure 1. Weighting structure under SDG target 2.2 on malnutrition.

Wasting_total
Wasting (weight=1/8)
(weight=1/4)
Wasting_girls
(weight=1/8)
Malnutrition
(Weight=1/2)
Overweight_total
(weight=1/8)
Target 2.2 Overweight
(weight=1) (weight=1/4)
Overweight_urban
(weight=1/8)
Stunting
(weight=1/2)

geographic location, or other characteristics. In 2018, UN on the overall progress depending on the progress of the
agencies conducted age, sex, and location disaggregation most vulnerable group in relation to the overall progress.
for more than 20 SDG indicators available in the global Figure 1 demonstrates a weighting structure to imple-
database. This number is expected to increase annually as ment this approach. In this scenario there are two indicators
global efforts to provide increasing granularity to official (malnutrition and stunting) under target 2.2, each receiving
statistics are becoming more commonplace. a weight factor equal to 0.5. The malnutrition indictor has
This section introduces modifications in current status and two types (wasting and overweight) for which disaggre-
anticipated progress gap indices that adjust for progress by gated values of prevalence are available for boys and girls
vulnerable groups.8 The idea is to introduce a multiplier fac- under-5, as well as across urban and rural population
tor that adjusts the overall progress made in one indicator groups. Those with higher prevalence are considered the
for the progress of the most vulnerable group. vulnerable group. For each type, vulnerable group is the
Assume that for each indicator I, there are two possible sub-population with the highest prevalence. For example, if
sets of sub-indicators; by type (e.g. prevalence of malnutri- prevalence of wasting is higher among girls than boys and
tion among children under-5 consists of two types, wasting higher in rural areas compared to urban, between rural pop-
and overweight) and by disaggregation variables (e.g. wast- ulation and girls, the one with the higher prevalence of
ing and overweight sub-indicators can be disaggregated by wasting should be considered the most vulnerable group.
sex and location). We mentioned earlier that all targets Likewise, if prevalence of overweight is the highest in urban
under each goal should be given equal importance, regard- areas, urban population is considered to be the most vulner-
less of the number of indicators. At the same time, all indi- able group. The weighting structure for this indicator is illus-
cators under the same target should also be given the same trated in Figure 1. For this particular target, one indicator
importance/weight regardless of number of sub-indicators. (stunting) and four sub-indicators (wasting_total, wasting_-
This means indicators and sub-indicators are weighted so girls, overweight_total, and overweight_urban) are used for
that the sum of weights under each target adds up to 1 assessing progress towards achieving the 2030 target. Sum
and indicators under each target are weighted equally. The of the weights is equal to 1 (4 9 1/8 + 1/2 = 1).
leave no-one behind aspect of the SDGs means that all sub- In ideal cases, it would be appropriate to have nested dis-
population groups should make required progress to aggregation, such as by age and sex. The following example
achieve the 2030 target. To account for this in compilation shows the weight structure for the indicator on proportion
of the progress index, one can multiply the overall progress of population living in extreme poverty, in which nested dis-
on each indicator by progress of the most vulnerable popu- aggregation is only available by employment, age and sex
lation group. This may have a negative or positive impact (Figure 2).

Figure 2. weighting model under SDG target 1.1 on extreme poverty.

extreme poverty in total


populaƟon
Target 1.1 extreme poverty (weight=1/2)
(Weight=1) (weight=1)
extreme poverty among
employed_male youth
(weight=1/2)

Global Policy (2020) 11:1 © 2019 University of Durham and John Wiley & Sons, Ltd.
Arman Bidarbakhtnia
62

In this example, in addition to disaggregation by employ-


Table 3. Comparison between current status and anticipated
ment, three types of nested disaggregation are available; progress indices before and after including vulnerable groups
employment and age (employed youth and employed
adult), employment and sex (employed male and employed Percentage change after including vulnerable
female) and employment, age and sex (employed youth by groups
sex and employed adult by sex). Among all nine subpopula- Current status index Anticipated progress gap
tions, extreme poverty is the highest among employed male Goal (%) index (%)
in youth age. Therefore, in addition to the main indicator
(extreme poverty in total population), extreme poverty 1 5 0
3 0 2
among identified vulnerable group (employed male aged
4 2 5
15–24) will also be incorporated in the progress assessment 6 25 7
under target 1.1. Having only one indicator under this target 8 33 0
and one vulnerable group, each will get a weight factor of
0.5.
Therefore, the number of data series (indicator, sub-indi-
cator or disaggregation) that can be used for assessing pro-
gress against one SDG target is n ¼ NS þ S þ V, where NS is vulnerable groups is as good as the rest of the population
number of indicators without any sub-indicator, S is total (zero values in Table 3), the percentage change in indices due
number of sub-indicators, and V is number of vulnerable to inclusion of vulnerable groups varies between 25 per cent
groups. In other words, V is number of indicators or sub- reduction (Goal 6) to 33 per cent increase (Goal 8). This means
indicators for which at least one disaggregation is available. that on average, since 2000, the most vulnerable groups have
Current status and anticipated progress gap indices intro- made faster progress than the rest of population in achieving
duced in formulas (1–3) must be constructed for each of the 2030 targets under SDG 6 (clean water and sanitation) and
n data series prior to aggregating at the target level. progress for the most vulnerable groups have been slower
Weighted average of current status and anticipated progress than the total population under SDG 8 (decent work and eco-
gap at goal level (see section ‘SDG progress assessment: nomic growth).
UNESCAP approach’) will be compiled as follows:

Pcs ¼
XX
k
ak  Pcs ðj ¼ 1; 2; ::; NÞ; ðk ¼ 1; 2; 3; ::; nÞ; ð5Þ Asia-pacific progress towards the SDGs in 2018–19
j k Figure 3 illustrates the results of applying the new approach
9 on data from all Asia-Pacific countries (when available) for
in which N is number of measurable targets under the goal,
k 17 SDGs. It combines the results on the current status in
ak is the weight of data series k, and Pcs is current status
2018 and the expected progress by 2030 in one graph. The
index compiled for data series k, and:
length of the bars shows the size of progress or regression
Ppg ¼
XX
k
ak  Ppg ðj ¼ 1; 2; ::; NÞ; ðk ¼ 1; 2; 3; ::; nÞ; ð6Þ of the region since 2000 against each of the goals. The
j k
length of each arrow is the size of progress or regression
we can expect by the end of 2030 if the region maintains
k the current pace of progress (business as usual).
in which Ppg is anticipated progress gap index compiled for
data series k: Data series for vulnerable group is given the
same weight as the same series for total population.
PP Sum of Acceleration of progress needed everywhere
the weights under each target has to be 1 ( ak ¼ 1). An
j k The results show that Asia-Pacific region will miss all 17 SDGs
alternative (and intuitive) way of compiling Pcs and Ppg with- by 2030 unless it accelerates the current pace of progress or
out using weights is to average data series in three separate reverses trends where it is regressing. Since 2000, the region
phases when relevant; first average over each sub-indicator has made better progress on zero poverty (Goal 1), quality
and its vulnerable groups, then over all sub-indicators of education (Goal 4), affordable and clean energy (Goal 7) and
each indicator, and finally average over all indicators. This is partnership for the goals (Goal 17). However, even on these
identical to the weighted sums in (6) and (7). goals, the region will miss the 2030 targets with the current
As of 2018, disaggregated values by employment, age, sex, speed. Sadly, the region has regressed on three SDGs, namely
and location were available only for 27 global SDG indicators clean water and sanitation (Goal 6), decent work and eco-
(out of 232) for more than half of the countries in Asia-Pacific nomic growth (Goal 8), and responsible consumption and
region10 (Appendix A). These indicators fall under goals on no production (Goal 12). If everything remains the same until
poverty (Goal 1), good health and well-being (Goal 3), quality 2030, Asia-Pacific is expected to make the greatest progress
education (Goal 4), clean water and sanitation (Goal 6), and in five goals; good health and well-being (Goal 3), quality edu-
decent work and economic growth (Goal 8). Table 3 illustrates cation (Goal 4), affordable and clean energy (Goal 7), industry,
a comparison of two indices (current status and anticipated innovation and infrastructure (Goal 9), reduced inequality
progress gap) before and after taking vulnerable groups into (Goal 10), and peace, justice and strong institutions (Goal 16)
account. Except three cases in which progress among and partnership for the goals (Goal 17).

© 2019 University of Durham and John Wiley & Sons, Ltd. Global Policy (2020) 11:1
Measuring the SDGs: An Inclusive Approach
63

Figure 3. Asia-Pacific SDG snapshot 2018–19: progress made so far and expected progress by 2030.

Progress made/ Progress/


regression since 2000 regression expected by 2030

Environmental degradation (see Appendix A)). Nearly two-thirds of global SDG indicators
cannot be used for assessing Asia-Pacific progress due to
The bad news is that judging from current trends, the
lack of data or methodology. Monitoring implementation of
region is expected to regress on sustainable cities and com-
the 2030 agenda will only be made possible by doubling
munities (Goal 11), climate action (Goal 13), life below water
efforts and investment in the production of timely and reli-
(Goal 14), and life on land (Goal 15) – all environmental. The
able disaggregated statistics.
good news is that, if the most recent trends continue, the
region will be able to reverse the current negative trends on
clean water and sanitation (Goal 6), decent work and eco- Conclusions
nomic growth (Goal 8), and responsible consumption and
More than ever, national statistical systems (NSSs) around
production (Goal 12) by 2030. Progress on Goal 12 is mainly
the world search for simple, comprehensive, and accurate
due to increased resource efficiency in production. However,
methods to measure success of their respective govern-
unsustainable consumption and emissions will continue to
ments in delivering national and international commitments
hamper the progress on this goal.
to sustainable development. The international community
has offered a range of approaches to monitor the complex
Traditional areas of social development development framework of the SDGs. This paper reviewed
Progress since 2000 shows that the region has put more three major approaches taken by SDSN, UNESCAP, and
emphasis on eradicating extreme poverty and improving OECD and highlighted their distinct features and commonal-
access to basic services such as education and health (Goals 1, ities. At the first glance, all three may look the same and
3, and 4). These are the three social goals from the MDGs era. serve the same purpose of ‘SDG progress assessment’, as
Until recently, the region had not prioritized reducing gender they are usually referred to. However, this detailed analysis
and other types of inequalities, providing access to justice for showed that three methods answer completely different
all, safe societies and strong institutions. Recent trends show questions. National statistical systems need to make impor-
(arrows) that the region has begun to accelerate progress in tant decisions prior to selection of their preferred measure;
reducing inequalities and improving peace, justice and strong what is the key question they aim to answer? What is their
institutions but it is too slow to meet the 2030 targets. basis of comparison? What serves their planning and prioriti-
zation the best? Table 1 provides a guide to select the most
relevant measure based on answers to the above questions.
Data remains a big challenge
Each method requires a different set of data, entails differ-
It is important to note that this analysis is based on only ent analysis and proceeds to a completely different interpre-
105 indicators (83 from global SDG indicators and 22 proxies tation. The term ‘progress made’ or ‘distance travelled’ could

Global Policy (2020) 11:1 © 2019 University of Durham and John Wiley & Sons, Ltd.
Arman Bidarbakhtnia
64

mean many different things depending on which measure is 10. For complete list of indicators, visit Asia-Pacific SDG gateway at:
used, what target is set, what comparison is made, how http://data.unescap.org/#item-2-7.
numbers are aggregated and how indicators are selected
and used. References
No measurement framework does justice to the 2030
agenda unless it incorporates disaggregated statistics and Bidarbakht-Nia, A. (2017a) ‘Tracking Progress towards the SDGs:
accounts for the progress among vulnerable groups across Measuring the Otherwise Ambiguous Progress’, UNESCAP Working
Paper SD/WP/05/May 2017.
all dimensions. This paper also introduces a modification to
Bidarbakht-Nia, A. (2017b) ‘A Weighted Extrapolation Method for
the UNESCAP measurement approach to make it inclusive of measuring the SDGs Progress’, UNESCAP Working Paper SD/WP/04/
progress by the furthest left behind groups. The results March 2017.
showed that inclusive indices can introduce up to 33 per Bidarbakht-Nia, A. (2018) ‘Regional Aggregates: Masking Change in
cent change in the progress. Regional Disparities?’, UNESCAP Working Paper SD/WP/06/March
The progress assessment for Asia-Pacific countries in 2018 2018.
Degol, H. and Raquel, T. (2011) ‘Achieving the Millennium Development
based on the new approach shows that the region needs to
Goals: A Measure of Progress’, International Policy Centre for
accelerate progress everywhere if it aims to achieve the Inclusive Growth Working Paper 78, February.
SDGs by 2030. With current trends, the region is expected Fukuda-Parr, S. and Greenstein, J. (2010) ‘How Should MDG
to regress on almost all environmental goals by 2030. Implementation Be Measured: Faster Progress or Meeting Targets?’,
International Policy Centre for Inclusive Growth Working Paper 63,
May.
Notes Kroll, C. (2015) Sustainable Development Goals – are the Rich Countries
Ready? Gutersloh: Bertelsmann Stiftung.
The author is grateful to the Statistical Data Management Nicolai, S., Hoy, C., Berliner, T. and Aedy, T. (2015) Projecting Progress:
team of UNESCAP for their contribution to the methodology Reaching the SDGs by 2030. London: Overseas Development Institute.
and data analysis. In particular, I am indebted to Dayyan NITI Aayog. (2018) SDG India Index: Baseline Report. New Delhi: NITI
Aayog.
Shayani for proofreading the early draft of the paper and
OECD. (2017) Measuring Distance to the SDG Targets 2017: An Assessment
for his excellent technical contribution to the concept and of Where OECD Countries Stand. Paris: OECD Publishing.
methods provided in this paper. The work underpinning this Osorio, R. G. (2008) ‘Alternatives for Projecting MDGs Indicators’,
research has gained immense support from senior manage- International Policy Centre for Inclusive Growth, Technical Paper 2.
ment of UNESCAP. However, views and opinions expressed Brasilia.
in this paper are those of the author and do not necessarily Sachs, J., Schmidt-Traub, G., Kroll, C., Durand-Delacre, D. and Teksoz, K.
reflect the position of any organization. (2016) SDG Index and Dashboards – Global Report. New York:
Bertelsmann Stiftung and Sustainable Development Solutions
Network (SDSN).
Sachs, J., Schmidt-Traub, G., Kroll, C., Durand-Delacre, D. and Teksoz, K.
1. https://unstats.un.org/sdgs/indicators/indicators-list/ (2017) SDG Index and Dashboards Report 2017. New York:
2. https://www.un.org/sustainabledevelopment/sustainable-developme Bertelsmann Stiftung and Sustainable Development Solutions
nt-goals/ Network (SDSN).
3. By progress assessment report I mean reports that allow measuring Sen, A. (1985) Commodities and Capabilities. Amsterdam: North Holland.
progress against goals/targets over time and not necessarily any Stanton, E. A. (2007) ‘The Human Development Index: A History’. PERI
report analyzing SDG indicators. Working Paper 127.
4. Median value of the indicator over all countries for which data is Tabatabai, H. (2007) ‘MDG Targets: Misunderstood or Misconceived?’,
available is used as the regional value (Bidarbakht-Nia, 2018). Med- International Policy Centre for Inclusive Growth, One Pager 33,
ian was preferred to weighted average to avoid bias towards bigger Brasilia.
economies or larger populations. The objective of this approach is UNDP. (2018) Human Development Indices and Indicators: 2018 Statistical
to give equal weight and equal responsibility to all countries Update. New York: United Nations Development Programme.
regardless of their size. UNESCAP/UNDP/ADB. (2003) A Future Within Reach: Reshaping
5. For indicators which the current value has already reached or Institutions in a Region of Disparities to Meet the Millennium
exceeded the target value current status index does not need to be Development Goals in Asia and the Pacific. Bangkok: United Nations
calculated and automatically is set to 10. Publications.
6. The weights should not be used for indicators that are time-inde- UNESCAP/UNDP/ADB. (2013) Asia-Pacific Aspirations: Perspectives for a
pendent such as disaster-related, ODA and other financial aids indi- post-2015 Development Agenda, August 2013, Bangkok: United
cator. Nations Publications.
7. It is crucial to remove outliers from data prior to identifying the top UNESCAP. (2017) Statistical Yearbook for Asia and the Pacific 2016: SDG
five performers. Baseline Report. Bangkok: United Nations Publications.
8. The proposed inclusive approach is applicable to any progress UNESCAP. (2018) SDG Data, Asia-Pacific SDG Gateway, updated on
assessment method. However, in this paper the adjustment is only September 2019. Available from: http://data.unescap.org/ [Accessed
introduced to the UNESCAP approach since it measures progress 06 December 2019].
over time (see Table 2). Vandermoortele, J. (2007) ‘MDGs: Misunderstood Targets?’, International
9. Targets for which at least one indicator is available. Policy Centre for Inclusive Growth, One Pager 28, Brasilia.

© 2019 University of Durham and John Wiley & Sons, Ltd. Global Policy (2020) 11:1
Measuring the SDGs: An Inclusive Approach
65

Appendix A. List of indicators used for SDG progress assessment in Asia and the Pacific region

.
Indicator

1.1.1 Proportion of employed population living on less than US$1.90 a day [total, 15–24 years and 25 + years, by sex]
1.1.1 Proportion of population living on less than $1.90 a day
1.2.1 Percentage of population living below the national poverty line [all area, rural]
1.5.1 Number of deaths, missing persons and directly affected persons attributed to disasters, per 100,000 population
1.5.2 Direct agriculture loss attributed to disasters, million US dollars
1.a.2 Proportion of total government spending on essential services [health and education]
2.1.1 Prevalence of undernourishment, percentage of population
2.2.1 Children moderately or severely stunted, percentage of children under 5
2.2.2 Prevalence of malnutrition, percentage of children under 5 [overweight and wasted]
2.3.P1 Cereal yield, kg per hectare
2.4.P1 Greenhouse gas (GHG) emissions from agriculture [per GDP from agriculture & CO2 equivalent]
2.5.2 Local breeds classified as being at unknown level of risk of extinction, percentage
2.a.1 Agriculture orientation index, Index
3.1.1 Maternal mortality, deaths per 100 000 live births
3.1.2 Births attended by skilled health personnel, percentage of live births
3.2.1 Under-five mortality rate, deaths per 1 000 live births [by sex]
3.2.2 Neonatal mortality rate, deaths per 1,000 live births
3.3.1 New HIV infections, per 100,000 population [All ages and 0–14 years, by sex]
3.3.2 Tuberculosis incidence rate, per 100,000 population
3.3.3 Malaria incidence rate, per 1,000 population at risk
3.4.1 Mortality rate attributed to cardiovascular disease, cancer, diabetes, or chronic respiratory diseases [by sex]
3.4.2 Suicide, per 100,000 population [by sex]
3.5.2 Alcohol per capita consumption, litres per annum
3.6.1 Road traffic deaths, per 100,000 population
3.7.1 Demand for family planning satisfied with modern methods, percentage of women of reproductive age
3.7.2 Adolescent fertility rate, live births per 1,000 women aged 15–19
3.8.2 Population with large household expenditure on health, percentage of population [more than 10per cent and 25per cent]
3.9.3 Mortality rate attributed to unintentional poisoning, Per 100 000 population [by sex]
3.b.1 Target population with access to vaccines, percentage of population [against DPT3, PCV3, MCV2]
3.c.1 Health worker density, per 1,000 population [dentistry, nursing and midwifery, pharmaceutical, physicians]
3.d.1 International Health Regulations (IHR) core capacity index, index
4.1.P1 Gross intake ratio to the last grade of lower secondary general education and to the last grade of primary education [by sex]
4.1.P2 Net enrolment in primary education, percentage of primary school age children
4.1.P3 Pupils enrolled who are at least 2 years over-age in lower secondary education and in primary education [by sex]
4.2.2 Participation rate in organized learning (one year before the official primary entry age) [by sex]
4.2.P1 Free pre-primary education guaranteed in legal frameworks, years
4.3.1 Proportion of 15- to 24-year-old enrolled in vocational secondary education, both sexes, percentage
4.3.P1 Gross enrolment in tertiary education, percentage of tertiary school-age population [by sex]
4.5.1 Gender parity for education indicators, female-to-male ratio
4.6.P1 Adult literacy rate, percentage of population aged 15 and above [by sex]
4.c.1 Trained teachers, Percentage [lower secondary, pre-primary, primary, secondary (by sex)]
5.1.P1 Gender wage gap, employees, percentage
5.1.P2 Labour force participation (aged 25+), female-to-male ratio
5.1.P3 Ratio of female to male mean years of schooling, population 25 + year, female-to-male ratio
5.1.P4 Youth labour force (15–24), female-to-male ratio
5.5.1 Seats held by women in national parliament, percentage of seats
5.5.2 Women share of employment in managerial position, percentage
5.5.P1 Women researchers, percentage of R&D headcount
6.1.1 Population using safely managed drinking water, percentage of population [by urban/rural]
6.2.1 Population practicing open defecation, percentage of population [by urban/rural]
6.4.2 Total freshwater withdrawal, percentage of total renewable water per annum
6.6.1 Water body extent (permanent), percentage of land area
7.1.1 Access to electricity, percentage of population
7.1.2 Population with primary reliance on clean fuels and technologies, percentage of population
7.2.1 Renewable energy consumption (SDG), percentage of total final energy consumption
7.3.1 Energy intensity (2011 PPP) (GTF), megajoules per unit of GDP in 2011 PPP
8.1.1 GDP per capita growth rate (2010 US dollars, average annual), percentage change per capita per annum
8.2.1 GDP per employed person, percentage change per annum

Global Policy (2020) 11:1 © 2019 University of Durham and John Wiley & Sons, Ltd.
Arman Bidarbakhtnia
66

.
Indicator

8.4.1 Material footprint, kg per 1 US$ (2010) GDP [total, biomass, fossil fuels, metal ores, non-metallic minerals]
8.4.2 Domestic material consumption, kg per 1 US$ (2010) GDP [total, biomass, fossil fuels, metal ores, non-metallic minerals]
8.5.2 Unemployment rate [15+, youth, by sex]
8.6.1 Not in employment, education, Ttraining (NEET), percentage of population aged 15–24 [by sex]
8.8.1 Frequency rates of fatal occupational injury, cases per year per 100 000 workers [by sex]
8.8.P1 Employees working more than 48 hours per week, percentage of employees [by sex]
8.10.1 Access to banking, insurance and financial service, per 100,000 adults [ATM and commercial banks]
8.10.2 Adults (15 years and older) with an account at a bank, percentage of population [by sex]
9.1.1 Air transport passengers carried, per 1000 population
9.2.1 GDP by activity: Manufacturing, percentage of GDP
9.2.2 Manufacturing employment (SDG), percentage of total employment
9.3.2 Proportion of small-scale industries with a loan or line of credit, percentage
9.4.1 Carbon dioxide (CO2) emissions per unit of manufacturing value added and from fuel combustion, kg per 1 US$ (2010) GDP
9.5.1 Gross domestic expenditure on research and development, percentage of GDP
9.5.2 Researchers, full-time equivalents, per million inhabitants
9.b.1 Medium and high-tech industry value added, percentage of total value added
9.c.1 Population covered by at least 2G, 3G and 4G mobile networks, percentage of population
10.1.P1 Gini index, income equality coefficient
10.2.1 Population living below 50 per cent of median income, percentage of population
10.4.1 Labour income share of GDP, percentage of GDP
10.c.1 Remittance cost as a proportion of the amount remitted, percentage
11.1.P1 Population practicing open defecation, percentage of urban population
11.2.P1 Road traffic deaths, per 100,000 population
11.5.1 Number of deaths, missing persons and directly affected persons attributed to disasters, per 100,000 population
11.6.P1 Annual mean concentration of PM2.5, micrograms per m3
12.2.1 Material footprint, kg per 1 US$ (2010) GDP [total, biomass, fossil fuels, metal ores, non-metallic minerals]
12.2.2 Domestic material consumption, kg per 1 US$ (2010) GDP [total, biomass, fossil fuels, metal ores, non-metallic minerals]
12.4.2 Hazardous waste generation, kg per capita
12.4.P1 Sulphur dioxide (SO2) emissions, kg per capita
13.1.1 Number of deaths, missing persons and directly affected persons attributed to disasters, Per 100,000 population
13.2.P1 Carbon dioxide (CO2) emissions from fuel combustion [per GDP & CO2 equivalent per capita]
13.2.P2 Greenhouse gas (GHG) emissions, total, metric tons of CO2 equivalent per capita
14.2.P1 Ocean health index, scores
14.5.1 Proportion of marine key biodiversity areas covered by protected area status, percentage
15.1.1 Forest area, percentage of land area
15.1.2 Important sites that are covered by protected areas, percentage [fresh water biodiversity & terrestrial biodiversity]
15.2.1 Progress towards sustainable forest management, percentage [forest area net change rate, forest area with a long-term
management plan, forest area within legally established protected area]
15.2.P1 Above ground biomass in forest, tons per hectare
15.4.1 Important sites for mountain biodiversity, percentage
15.5.1 Red list index total, index
16.1.1 Intentional homicide, per 100,000 population
16.3.2 Unsentenced detainees (pre-trial), percentage of prison population
16.b.P1 Internally displaced persons, thousand people
16.b.P2 Refugees by country of origin, thousand people
17.1.1 Tax revenue (GFS-G11), percentage of GDP
17.3.2 Personal remittances received (LDCs), percentage of GDP
17.3.P1 FDI inflows (LDCs), percentage of GDP
17.4.1 Debt service, percentage of exports of goods, services and primary income
17.6.2 Fixed-broadband equal to or above 10 Mbit/s subscriptions, per 100 population
17.8.1 Internet users, percentage of population
17.9.1 Official development assistance (gross disbursement) for technical cooperation, million 2016 US dollars
17.10.1 Tariff rate for LDCs (WITS) under most favoured nation for manufactured and primary products (LDCs), percentage
17.11.1 Exports from LDCs for commercial services and merchandise, percentage of world services exports
17.19.1 Resources made available to strengthen statistical capacities in developing countries, million US dollars
17.19.P1 ODA to statistical capacity building (CRS 16062), million 2015 US dollars

© 2019 University of Durham and John Wiley & Sons, Ltd. Global Policy (2020) 11:1
Measuring the SDGs: An Inclusive Approach
67

survey methodology, consumer demand analysis and measuring human


Author Information development. He has previously worked as survey statistician at the Sta-
Arman Bidarbakhtnia is statistician at the United Nations (UN). He tistical Centre of Iran and as lecturer at the UN Statistical Institute for
holds a PhD in welfare economics and has published his research on Asia and the Pacific (SIAP).

Global Policy (2020) 11:1 © 2019 University of Durham and John Wiley & Sons, Ltd.

You might also like