qualitative or quantitative research

Socio-Economic Planning Sciences 79 (2022) 101101

Available online 25 June 2021 0038-0121/© 2021 Elsevier Ltd. All rights reserved.

A quantitative approach for analysis of macroeconomic resilience due to socio-economic shocks

Hojat Rezaei Soufi, PhD Candidate, Akbar Esfahanipour, Associate Professor *, Mohsen Akbarpour Shirazi, Associate Professor Department of Industrial Engineering and Management Systems, Amirkabir University of Technology, Tehran, Iran

A R T I C L E I N F O

Keywords: Macroeconomic resilience Socio-economic shocks COVID-19 pandemic Data envelopment analysis DEMATEL

A B S T R A C T

Macroeconomics has constantly been exposed to socio-economic shocks. The concept of resilience in the econ- omy has been developed to predict these shocks, reduce damages, and recover quickly. This paper proposes a quantitative approach for analyzing macroeconomic resilience due to socio-economic shocks and suggests appropriate actions to improve resilience. In this way, the variables affecting macroeconomic resilience have been identified through the literature review. Next, an integrated indicator of macroeconomic performance based on Data Envelopment Analysis (DEA) has been developed. Finally, the periods of shocks are identified by determining the turning points of that indicator, and an appropriate approach for defining macroeconomic resilience is developed. The proposed approach is applied to three countries of the USA, China, and Iran in different shocks, including global crisis, COVID-19 pandemic, and oil price shock. Eventually, by analyzing the relationships between effective variables on the macroeconomic resilience, using the DEMATEL method, we determine the most important variables to improve macroeconomic resilience, which can be useful for socio- economic planning at a macro level.

1. Introduction

In recent years, the issue of economic resilience, especially after the 2008 financial crisis, has attracted many researchers. Recent events, such as the COVID-19 pandemic, have also challenged the resilience of economies. According to Briguglio et al. [1], economic resilience is the capability of an economy to avoid economic shocks and rapid recovery to main functionality. This definition refers to socio-economic shock, which is an unexpected event that has a large-scale and unexpected impact on the economy [2].

Investigating the resilience of the macroeconomy and its effective variables provides a basis for enhancing this concept and improving the security and stability of the economy against various crises [3]. Halle- gatte [4] defines macroeconomic resilience as the value of the lost asset of an economy during a disaster. In his viewpoint, resilience is related to the reduced functionality and the required time to recover the econo- my’s functionality to a normal level. Therefore, to examine macroeco- nomic resilience, we need to determine the effective variables on macroeconomic resilience [5] and develop an integrated indicator for macroeconomic performance from the resilience viewpoint.

Furthermore, the relations between the variables should be considered to consider macroeconomic resilience improvement. It is notable; some variables have a maximizing nature, which means that their higher value is good, and some variables have the opposite nature. In some cases, variables have a mixed nature, and it depends on other variables.

The main challenges of this study are:

• Developing an indicator showing the macroeconomic performance from the resilience viewpoint;

• Selecting the variables with the ability to express the macroeconomic at specified time intervals accurately;

• Determining the nature of variables, the relation between them; • Integrating the variables and developing a macroeconomic perfor-

mance indicator; • Identifying turning points in macroeconomic performance to deter-

mine the socio-economic shock period; • Developing methods to measure macroeconomic resilience; • Examining variables to promote resilience as a supportive approach

to decision making.

* Corresponding author. No. 350, Hafez Ave, Valiasr Square, Tehran, 1591634311 Iran. E-mail addresses: h.rsoufi@aut.ac.ir (H. Rezaei Soufi), esfahaa@aut.ac.ir (A. Esfahanipour), akbarpour@aut.ac.ir (M. Akbarpour Shirazi).

Contents lists available at ScienceDirect

Socio-Economic Planning Sciences

journal homepage: www.elsevier.com/locate/seps

https://doi.org/10.1016/j.seps.2021.101101 Received 11 August 2020; Received in revised form 6 May 2021; Accepted 21 June 2021

Socio-Economic Planning Sciences 79 (2022) 101101

2

To this point, first, through a literature review, the relevant variables are identified. Next, the variables which can show the status of the macroeconomic in a specified period are separated. It is notable; the resilience indicator should have the ability of traceability through time [6]. Therefore, the selected indicators should not have a long reporting period. It should also provide accurate information about macroeco- nomic. Therefore, it is better to avoid using qualitative variables. The next step is to integrate these variables by applying Data Envelopment Analysis (DEA) method. The final step is developing different methods for measuring macroeconomic resilience. As a complementary step, by examining the data correlation and analyzing the relationship between the variables, a decision-making procedure for improving resilience is developed.

In the paper, various socio-economic shocks and their effects on the studied countries’ macroeconomics are examined. Recently, with the spreading of the COVID-19 pandemic, studying macroeconomic resil- ience due to the coronavirus has become very important, and several studies are being conducted in this field [7]. Therefore, in this paper, the macroeconomic resilience of the pandemic has been analyzed by extending of our data into the post-coronavirus period.

2. Literature review

In recent years, there are more accurate definitions of macroeco- nomic resilience. Resilience is the ability of a system in its function of duty (such as production continuity) when a shock occurs [8]. This definition is based on the fundamental problem of the economy, which is the allocation of scarce resources. A more general description incorpo- rating dynamic considerations and can bring the speed at which a system is recovering from a severe shock is called a dynamic economy [9,10]. Another definition is the ability to absorb losses or to improve rapidly [11]. Kamissoko et al. [12] define resilience as the ability to maintain the duty of a system in critical situations through developing the skills and making more efforts, for example, increasing the possibilities of success in commercial operations or strengthening the market by providing information to coordinate the breadth of the recipients and customers.

Due to the global financial crisis in recent years, the economic resilience concept is developed. The researchers develop different defi- nitions and various methods for economic resilience and provide a different list of indicators for economic resilience.

Brigogulio et al. [13] suggest that the simple resiliency index is per capita GDP because this variable encompasses the country’s ability to deal with vulnerabilities. Brigogulio [1] presents the first indicator of an economic downturn. In his view, he respects at least three potentials in a single economy: the economy’s ability to avoid these shocks, the ability of an economy to withstand the effects of these shocks, and the ability of an economy to recover quickly from external economic shocks. Brigo- gulio et al. [1] also believe that a low unemployment rate, low inflation rate, lower foreign debt ratio, and Public debt to GDP ratio make an economy more resilient. Furthermore, in his viewpoint, the small-sized governments have a more resilient economy with lower debt and lower state ownership. He considers stability and flexibility as two main components of resiliency and presents a list of indicators for economic resilience. Boorman et al. [14] develop a study for evaluating the resilience of emerging markets and developing countries (EMDCs). They categorize economic resilience indicators in Fiscal Policy Government effectiveness, Governance, Monetary Policy, Banking Soundness, Export diversity, Export independence, external robustness, Private debt, and Reserves. UK Asset and Wealth Management (AWM) strategy team studies about community economic resilience index and classifies them in the macroeconomic, labor market, and social classes. Boorman [15] analyzes 52 different variables to assess the ability to develop an indi- cator for EMDCs to deal with shocks; they group these variables into ten sub-indicators. The most important variables are financial policy health, including Public debt to GDP index and its rate of change. The second is

Monetary Policy Health, including the difference between domestic inflation and inflation in G7 countries. The third is Government Effec- tiveness, which shows the ability of governments to respond to shocks. The fourth is general governance, including the rule of law, trans- parency of dealing with corruption, freedom of the press. Other vari- ables are the banking system’s health, the variety of exports, Export dependence, external power, Private sector debt, and the net investment status of international and international reserves.

Angulo et al. [16] use the employment rate to evaluate resilience to socio-economic shocks. They develop two different quantitative mech- anisms to calculate the economic resilience, including Adaptive (i.e., the traditional shift-share in employment rate) and engineering/ecological (i.e., the path of employment rate during pre-and post-crisis periods). Rohn et al. [17] define resilience as a tool for minimizing potential vulnerabilities in coping with external events and propose a list of vulnerability indicators for OECD countries. In their viewpoint, the potential vulnerabilities are related to both domestic and international sides. Their proposed parameters are the financial sector, Non-financial sector, Asset markets, Public sector, external sector, and the foreign sector. Mirzaei and Al-Khouri [18] analyze Kuwait’s economics’ resil- ience as an oil-rich country to the 2007 global financial crisis. They investigate the bank performance and industry growth during the shock period and develop different regression models to check it. They find that Kuwaiti banks were negatively affected by the crisis and a shift in industries’ performance. They believed that these results would cover existing weaknesses and promote the resilience of oil-exporting eco- nomics. Marto et al. [19] analyze the macroeconomic impact of major natural shocks. They consider different macroeconomic variables before and after the shocks at different levels (firms, government, and households).

The studies mentioned above provide assumptions about the rela- tionship between economic resilience and these variables by developing different models. Hallegatte [4], for the first time, introduces a function for calculating economic resilience. Indeed, his developed function calculates the value of lost assets after a disaster in an economy. His proposed method also contributes a number to economic resilience; however, the technique has several challenges with the real world. First, the decrease in productivity happens all at once. Second, the level of productivity fixes during a shock period. Finally, the productivity returns to normal conditions. Therefore, this is better to improve their developed model for considering lost assets as an indicator of economic resilience to have a more realistic model.

Other approaches developed to analyze the economy’s resilience are input-output-based models that analyze the overall resilience by iden- tifying different sub-sectors of an economy and analyzing the risks of each component. Rose [20], in a study on economic resilience due to earthquakes, categorizes resilience proceedings into inherent (actions in normal and preparatory conditions) and adaptive (actions in critical conditions) and believes that economic resilience can be studied in three areas: micro, meso, and macro. He believes that the economy’s resil- ience was the result of correct adaptive plans in the micro and meso sectors and inherent plans in the macro sector. In another study, Pant et al. [21] calculate economic resilience in both static and dynamic states. They use input-output models to develop an indicator for the economy and examine resilience concerning independent infrastructure and industry sectors. A static approach based on the rate of performance change and a dynamic approach based on the concepts of performance change and recovery time were established in their study.

A review of the literature on macroeconomic resilience is contro- versial in some respects. First, identified indicators have been very various. Second, the indicators are not such that they can show the macroeconomic performance in short time units which is very important in calculating resilience [22]. Third, the integration of the related var- iables for the indicators has not been done accurately. The integrated indicator is not accurate in describing the macroeconomic status and is limited to a single type of categorization. The identified variables have

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

3

different natures (maximizing, minimizing, or mix nature). Also, there are no precise mechanisms for promoting resilience, providing the conditions for policymakers to make decisions. These issues can be considered as gaps in this area.

Therefore, this study aims to cover these gaps and develop a quan- titative approach for measuring macroeconomic resilience. In this way, the relationships between variables are analyzed, and in a coherent model, the functional measure of the macroeconomic will be obtained from the resilience perspective, and the quantitative macroeconomic resilience is measured. The model’s performance in various shocks, including the COVID-19 pandemic, is examined in the following. Finally, using the DEMATEL method, an analysis of the relationships between these variables is performed to improve resiliency.

Accordingly, the main contributions of this paper are as follows:

• Analyzing the macroeconomic variables and their nature in terms of maximizing. Minimizing or mixed;

• Integrating the variables to develop an indicator with the ability to express the macroeconomic status in specific time units quantitatively;

• Determining the socio-economic shocks’ periods through an appro- priate algorithm;

• Developing an indicator to measure the resilience of macroeconomic; • Investigating the performance of developed indicators by analyzing

the relations between the effective variables to develop a decision support structure.

Table 1 presents the most important features of this study compared to other similar studies in this field.

This study has been organized based on the following sections: The study’s proposed approach will be described accurately in the next section. It includes determining relevant variables, integrating the var- iables to develop an indicator, determining socio-economic shocks’ pe- riods, measuring the macroeconomic resilience, and analyzing the developed measure approach. In the next section, the data are described, and the results have been presented. In the fourth section, the results of the study are presented in different countries. In the fifth section, the findings obtained from the data are discussed, and in the sixth section, the conclusions derived from the study will be presented.

3. Our proposed approach

The approach of this study includes four main steps as 1) developing macroeconomic resilience indicator, 2) developing an aggregate func- tion for the macroeconomic indicator, 3) measuring macroeconomic resilience, 4) Determining the socio-economic shocks’ periods, and 5) studying the relationship between the variables. Fig. 1 presents the flowchart of the proposed approach.

3.1. Variables’ selection

In order to evaluate the macroeconomic performance during different shocks, a set of macroeconomic variables with available monthly data addressing in the literature are used. Next, the integrating process is applied. The main variables of the study are identified by reviewing the literature. However, the resilience indicator’s nature should have the capability to show the system’s state in shorter periods. There are several challenges in this section. The first challenge is different reporting time intervals for the variables such as annual, sea- sonal, monthly, and even shorter ones. The second is to select the var- iables having quantitative nature with certain or uncertain amounts.

Hence to increase accuracy in calculations, we will develop a quanti- tative indicator of the reporting approach in the shortest possible period by maintaining the indicator’s integrity. Accordingly, in Table 2, we report the variables which can be used in our study. In order to examine the nature of the variables in the maximizing or minimizing terms, we use the developed variables for macroeconomics addressed in the annual reports of the IMF, the World Bank, and the Organization for Economic Co-operation Development [6,15,17,26]. The results are presented in the second column of Table 2. According to the table, some variables have maximizing nature, several have minimizing character, and the others have a mixed nature.

3.2. Integrating the variables with DEA method

The selected macroeconomic variables should aggregate in a model to show the macroeconomic situation to evaluate the macroeconomic resilience during different shocks. There are different approaches to aggregating. For example, Cisse and Barret [27] used a modified weighted sum function to calculate the economic resilience obtained for a set of households. Mohanty and Sahoo [25]’s approach applied to aggregate macroeconomic variables in an integrated measure. They use DEA to calculate macroeconomic performance in India. They first normalize each macroeconomic variable and attribute them to a number between zero and one so that zero shows the worst performance, and one shows the best performance in the whole period of study. The standard DEA model tries to maximize the efficiency of decision-making units (DMUs). To this point, we consider DMUs as the macroeconomic vari- ables in different months. In the DEA models, each DMU is assigned to the best weights to evaluate the relative efficiency, calculated in the model [28]. The DEA model tries to maximize the efficiency of macro- economic indicators weighting in different periods. To calculate the value of macroeconomic performance (MEP) for each country in each time step, we use Sahoo and Acharya [24]’s non-radial DEA model DEA models by forming an efficient boundary to try to distinguish between efficient and inefficient DMUs. In the radial model, the inefficient DMUs by reducing inputs or increasing outputs can be depicted in an efficient area. However, non-radial models are based on the amount of slack and try to improve efficiency. The movement is parallel to the inlet and outlet axes and along the effective boundary in these methods. They apply a non-radial slack-based measure (SBM) DEA method to calculate the MEP of 22 Indian states.

The main challenge of Sahoo and Acharya [24]’s approach is to ignore the exact nature of the variables. The variables may have a maximizing, minimizing, or mixed nature, which is considered in this paper. With this categorization, the selected ten variables become fourteen variables with four maximizing and ten minimizing variables. Equations (1)–(14) show relations to calculate the normalized value of the variables. Note that in these relationships, each variable’s maximum and minimum values are the largest and smallest values observed for each case study over a predefined period. The normalization of these variables varies depending on whether they are maximizing or minimizing.

PDtGDP− nt = PDtGDPmax − PDtGDPt

PDtGDPmax − PDtGDPmin (1)

NPLtTL− nt = NPLtTLmax − NPLtTLt

NPLtTLmax − NPLtTLmin (2)

EtTL− nt = EtTLt − EtTLmin

EtTLmax − EtTLmin (3)

H. Rezaei Soufi et al.

Socio-EconomicPlanningSciences79 (2022)101101

4

Table 1 Features of our study in comparison with the related literature.

Reference Effective Variables Method Period

Fiscal Policy Soundness

Government effectiveness

Monetary policy

Corporate Governance

Legal Asset Quality

Capital Base

Risk Export diversity

External robustness

Private Debts

Labor Market

Social

Boorman et al. [14]

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Logistic regression

annual

Anulo et al. [16] ✓ ✓ ✓ ✓ ✓ ✓ ✓ – annual AWM strategy

team [23] ✓ ✓ ✓ ✓ ✓ ✓ ✓ – annual

Brigogulio [1] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ – annual Mirzaei and Al-

Khouri [18] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Logistic

regression Seasonal

Rohn et al. [17] ✓ ✓ ✓ ✓ ✓ Logistic regression

annual

Hallegatte [4] ✓ Resilience function and regression

Seasonal

Brigogulio et al. [13]

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ – annual

Boorman [15] ✓ ✓ ✓ ✓ ✓ ✓ regression Seasonal Sahoo, and

Acharya [24], ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ regression annual

Mohanty and Sahoo [25]

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Logistic regression

annual

Kammissoko et al. [12]

✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ _ annual

Rose [9] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ – Rose [20] ✓ ✓ Input-output annual Pant et al. [21] ✓ ✓ ✓ Input-output,

Resilience function

annual

Presented study ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Resilience function

monthly

H . R

ezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

5

SMR− nt = SMRt − SMRmin

SMRmax − SMRmin (4)

EXtGDP− nt = EXtGDPt − EXtGDPmin

EXtGDPmax − EXtGDPmin (5)

CPI+t = CPIt − CPImin

CPImax − CPImin (6)

PPI+n t = PPIt − PPImin

PPImax − PPImin (7)

UER−t = UERmax − UERt

UERmax − UERmin (8)

ExR−t = ExRmax − ExRt

ExRmax − ExRmin (9)

ItR−t = ItRmax − ItRt

ItRmax − ItRmin (10)

CPI−t = CPImax − CPIt

CPImax − CPImin (11)

PPI− n t = PPImax − PPIt

PPImax − PPImin (12)

ExR+t = ExRt − ExRmin

ExRmax − ExRmin (13)

ItR+t = ItRt − ItRmin

ItRmax − ItRmin (14)

Our proposed DEA model is presented in Equation (15), which is described.

Fig. 1. The flowchart of our proposed study.

Table 2 The selected quantitative variables for macroeconomic functionality.

Variable Variable nature

Reporting Period

Public debt/GDP (PDtGDP) Minimizing Seasonal Bank Nonperforming Loans to Total Loan

(NPLtTL) Minimizing Seasonal

Equity to the total asset (EtTL) Maximizing Seasonal Stock market return (SMR) Maximizing Daily Exports/GDP (EXtGDP) Maximizing Seasonal Consumer price index (CPI) Mixed nature Monthly Producer price index (PPI) Mixed nature Monthly Unemployment rate (UER) Minimizing Monthly Exchange rate (ER) Mixed nature Daily Interest rate (ItR) Mixed nature Monthly

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

6

where t is the index of time periods, j is the index of macroeconomic variable, i is the index of country and si s are the slacks in normalized macroeconomic variables, and λjs are the intensity coefficients which are interpreted as shadow prices. Note that state “h” is efficient if MEPh = 1 and MEPh = 1 if the slacks are zero. Also - and + signs are related to variables nature (maximizing or minimizing). The efficiency value in this model is between zero and one, taking the value of one when all the auxiliary variables are zero. If any of these slacks get a maximizing number, the state “h” is inefficient. Furthermore, to calculate the MEP of each country in each time step, the model should be run for T*J times where T is the total number of periods and J is the total number of variables.

3.3. Measuring macroeconomic resilience

According to the resilience definition, the proposed model should consider the effect of a shock in functionality decreasing and recovery time. This model has a dynamic structure by simultaneously considering the rate of performance reduction and recovery status at specific times while examining the percentage change in status can only have a static view of resilience.

Our proposed approach to measure macroeconomic resilience is based on Fig. 2, in which the highlighted area is the total loss of mac- roeconomic resilience (LOMeR). This area is related to two variables of the recovery time and the decreased level of MEP. Zobel and Baghersad use this structure to study the resilience of a human system during socio- economic shocks [29]. Determining the starting and endpoints of the crisis is important in calculating resilience using Fig. 2. Lucija [30] introduced five situations to describe the resilience of a system in socio-economic shocks: resistance (no change in status), recovered (full return to pre-crisis level), recovered but again downturn (second stage of shock), not recovered but in the upturn, not recovered but on the downturn. Here because the index is not returned to its previous value fully, the existence of a time lag from the onset of shock until the

downtrend begins, and the high fluctuation of the index, we use the turning point concept. Hence, the start and end of the crisis period are the turning points of the MEP pattern (see the next section). The decreased level of MEP is the level of reduction after a shock. Therefore, a lower level of reducing functionality and a shorter time to recovery leads to a higher level of resilience. Accordingly, we can quantify our definition of macroeconomic resilience as equation (16).

LOMeR = ∫ t2

t1 MEP(t)dt (16)

3.3.1. Detecting the turning points In order to calculate the resilience, it is essential first to identify the

starting point of the shock and starting point of resuming after the shock. These points are called turning points, as shown in Fig. 2. According to Yin et al. [31], turning points are the points of transforming an increasing trend to a decreasing trend or transforming a decreasing trend to an increasing trend. The process of converting from an ascending trend to a descending trend or vice versa should be considered

Fig. 2. A schematic view of the loss of macroeconomic resilience (LOMeR) function.

(MEPh) − 1

= max 1 + 1 14

( sPDtGDP

n

PDtGDPnh +

sNPLtTL n

NPLtT Lnh +

sEtTL n

EtT Lnh +

sSMR n

SMRnh +

sExtGDP n

ExtGDPnh +

sCPI +n

CPI+nh +

sCPI − n

CPI− nh +

sPPI +n

PPI+nh +

sPPI − n

PPI− nh +

sUER n

UERnh +

sExR − n

ExR− nh +

sExR +n

ExR+nh +

sItR − n

ItR− nh +

sItR +n

ItR+nh

)

for each t and j S.t. PDtGDPnj λ

t j − s

PDtGDPn = PDtGDPnth ∨ t,i

NPLtT Lnj λ t j − s

NPLtTLn = NPLtTLnth ∨ t,i

EtT Lnj λ t j − s

EtTLn = EtT Lnth ∨ t, i

SMRnj λ t j − s

SMRn = SMRnth ∨ t, i

ExtGDPnλtj − s ExtGDPn

= ExtGDPnth ∨ t,i

CPI+nj λ +t j − s

CPI+n = CPI+nth ∨ t, i

CPI− nj λ − t j − s

CPI− n = CPI− nth ∨ t, i

PPI+nj λ +t j − s

PPI+n = PPI+nth ∨ t,i

PPI nj λ − t j − s

PPI− n = PPI− nth ∨ t,i

UERnj λ t j − s

UERn = UERnth ∨ t, i

ExR+nj λ +t j − s

ExR+n = ExR+nth ∨ t, i

ExR− nj λ − t j − s

ExR− n = ExR− nth ∨ t, i

ItR+nj λ +t j − s

ItR+n = ItR+nth ∨ t, i

ItR− nj λ − t j − s

ItR− n = ItR− nth ∨ t, i

∑14

j=1 λtj + λ

+t j + λ

− t j = 1 ∨ t, i (15)

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

7

to determine these points. These changes can be in the form of a sharp drop (rise) or a slow downward (upward) trend. To this end, Algorithms 1 and 2 are developed to determine the starting point of the shock period (t− ) and the starting point of the return period (t+). According to algo- rithm 1, the shock period starts if the reduction in the value of MEP is greater than a threshold equals to the average of negative returns minus twice of its the standard deviation in a specific period. Otherwise, if there are two consecutive time periods with negative returns (i.e., changes), and the sum of them are greater than that of the new threshold, which is calculated by average negative returns and its standard deviation with a coefficient in different lags, the shock period starts in the first lag which passed the limitation. Likewise, for more consecutive time periods (from the time i to j lag after it), if the sum of consecutive negative returns is greater than the new threshold value, the t- is identified. There is a similar pattern in the return phase, as defined in algorithm 2. Note that in this algorithm, to consider both sharp drop (rise) or slow downward (upward) modes, the algorithm is designed to consider only a heavy fall (rise) at the beginning (end) of the crisis as the t− (t+), and over time, negative (positive) returns (albeit small) can be considered to determine the t− (t+). Indeed, in the primitive stages, only long jumps have a chance of passing the threshold value. With a sequence of negative or positive returns, this threshold value does not lead to defining turning points.

3.4. Promoting resilience as a supportive approach to decision making

To this point, here, we study the relationship between variables with the DEMATEL method. In order to determine the effects of variables on each other, this study applies the DEMATEL method. This technique uses the experts’ views and defines the factors as nodes and relationships between the factors as arcs; then, it creates a network of direct and in- direct relationships between them [32]. Due to this method’s high ef- ficiency and accuracy, the study uses the DEMATEL method to determine the exact structure of the variable’s relationship network [33]. The steps of the DEMATEL method used in this study are as follows:

Step 1- Gathering information from the effect of variables on each other to create the initial matrix of relationships between the variables. The common approach in this step is to use the expert’s opinions on the effect of variables on each other. However, since the issue of using expert opinions and qualitative comparisons will reduce the calculation accuracy, this paper gets the advantage of using the Conditional Value at Risk (CoVaR) measure as a new approach to constructing the initial matrix. The CoVaR indicates the impact of change in one factor resulting from a change in another factor [34]. In this way, the issue of impacting the indicators on each other is considered with this measure. In this way, the issue of impacting the indicators on each other is considered with this measure.1 Accordingly, the matrix will be constructed as shown in (17), where CoVaR12 is the conditional value at risk of variable 2 when the first variable with a specified confidence level is less than its value at risk (VaR2).

A =

⎡

⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣

− CoVaR21 CoVaR 3 1 … CoVaR

N 1

CoVaR12 − CoVaR 3 2 … CoVaR

N 2

: : − … : : : CoVaRji − : CoVaR1N CoVaR

2 N … … −

⎤

⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦

(17)

Step 2- Calculating the normal matrix of direct relationships (D): To

normalize the matrix calculated in the previous step, we must apply the following transformation and calculate the matrix D. Note that aij will be the initial matrix elements in the previous step. Let s as the normaliza- tion factor; then the normalized matrix will be calculated as follows:

s = min

⎡

⎣ 1

max i

∑n j=1

⃒ ⃒CoVaRij

⃒ ⃒ ،

1 max

j

∑n i=1

⃒ ⃒CoVaRij

⃒ ⃒

⎤

⎦ (18)

1 Readers can refer to [35] to read more about CoVaR. 2 The estimate of the loss of a financial variable with a given probability.

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

8

D = S*A (19)

Step 3 - Calculating the matrix of total relationships (T): The matrix T, which is indicative of the total (direct or indirect) effects of factors on each other, is calculated using equation (20). The elements of this matrix will be tij, which reflect the sum of total effects of factor i on factor

T = D(I − D)− 1 (20)

Step 4: Determining the influence and final rate of the variable. In this step, considering the results obtained from the previous step,

the relationship network between factors can be depicted. Afterward, we need to determine the degree of influence and the degree of relationship of each node in the network. In fact, the influence rate of each variable on the other variable will be calculated here. The overall effects calcu- lated in this step will constitute the same synergistic factor of each variable we were willing to obtain. Mangla et al. [36] consider the sum of elements of ith row or the jth column of matrix T as influence rate of factor i (r)3 and dependency rate of factor i (c)4 as shown in equations (21) and (22).

ri = ∑n

j=1 tij (21)

cj = ∑n

i=1 tij (22)

According to Mangla et al. [36], ri + ci is called prominence for each factor that shows the significance of variable i between the others and represents the total effect of variable i and ri-ci are called relation for each factor that shows the total effect of variable i on other variables.5

When applying the DEMATEL method, researchers use different approaches to prioritize the variables. Some of them use the r + c, and others use the r-c measure and each of which emphasizes on the benefits of their measure in different situations (c greater than r or vice versa). One of the concepts that can be used in this section is the synergy measure, which is about how changing a variable can enhance the sys- tem. In fact, this concept simultaneously takes into account both the prominence of one indicator over the other indicators and the relation between them. In this paper, using the following concept, combining prominence and relation degree, we determine the synergy.

Hence, we develop a function, including r-c (the severity degree) and r + c (the relation degree), and normalize them to calculate an appli- cable measure. Moreover, to consider the preference of prominence and relation in the model, we define a preference coefficient (Gama) parameter. In this study, the Gama coefficient is calculated by applying the Entropy method. The developed synergy effect by this study is as equation (24).

SEi = γ (

ri + ci ∑n

i=1ri + ci

)

+

(

1 − γ )(

ri − ci ∑n

i=1 ri − ci

)

(24)

4. Data analysis and results

In this section, we use the relevant data of the United States, China, and Iran to evaluate their macroeconomic resilience. To show the applicability of our proposed approach in various conditions, we use these three countries’ data since they are in different macroeconomic situations. In this manner, the relevant data are gathered from 2006 to 2017. Then the data are integrated to calculate the MEP. Next, the crisis periods have been determined through our proposed algorithms in section 2-4, and finally, the resilience of each macroeconomic are calculated. It is notably the data source for gathering data is the

“tradingeconomics” website. The ratios mentioned in Table 2 are calculated as economic indicators for three countries of the US, China, and Iran in the same way. For example, general debt, GDP, bank non- performing loans, and total loans are collected separately, and the ra- tios of public debt to GDP and bank non-performing loans to total loans are calculated. Similarly, the other rates are calculated.

4.1. MEP calculation

Using the proposed DEA approach in section 2-2, the monthly MEP is calculated. The results are presented in appendix A. Figs. 3–5 show the relevant results for the US, China, and Iran, respectively.

The shocks’ periods are defined using the proposed algorithms for identifying the turning point in section 4-2. Results are shown in Table 3, which shows five shock periods for the US, three shock periods for

Fig. 3. Macroeconomic performance (MEP) for the United States.

Fig. 4. Macroeconomic performance (MEP) for China.

Fig. 5. Macroeconomic performance (MEP) for Iran.

3 How much does this variable affect the other variables altogether?. 4 How much the other variables collectively affect this variable?. 5 When (ri–cj) is positive, the variable is considered as a cause variable, and

when it is negative, it is considered as an affected variable.

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

9

China, and two shock periods for Iran.

4.2. Measuring macroeconomic resilience

According to the identified turning point in Table 3 and equation (16), the value of LOMeR for the United States, China, and Iran are presented in Table 4.

According to Table 4, the worst resilience is for the US in the 2008 global financial crisis. Furthermore, the US’s resilience during the 2014 oil price crisis is better than that of Iran and China as two oil suppliers. Moreover, the resilience of China during China’s stock market shock does not have a good situation. More analysis of these results is pre- sented in the next section.

5. Discussion

This section is about the determination of the most important vari- ables affecting macroeconomic resilience that countries should pay particular attention to them as well as discovering the relationships and impacts between the variables by using the DEMTEL method. Further- more, considering the importance of the COVID-19 pandemic case, data have been extended to investigate this shock, and the macroeconomic

resilience of the three understudied countries in this shock has been analyzed.

The initial matrix was computed to apply the DEMATEL method by calculating the CoVaR between the variables and then using the calcu- lations process presented in Section 2-4 normalized matrix, the final direct effects matrix, and the final direct and indirect effects matrix were obtained. Then, by calculating the influence rate of factor and de- pendency rate of factor (ri and ci) results, the final results are obtained.

The results of the weights of the variables for the US, China, and Iran are presented in Tables 5–7, respectively.

5.1. Analysis of results

In this step, we analyze the sensitivity of LOMeR to the indicators which is identified in this study. Accordingly, we change each indicator by +5% and calculate the LOMeR in each crisis period and the total duration of data. Table 6 shows the results. Furthermore, the relevant figures are presented in Figs. 6–8.

Table 8 shows that for the US, a 5% positive change in public debt to GDP ratio (PDtGDP) has the best effect on LOMeR and increased resil- iency by up to 8%. Unemployment rate (UER), equity to total asset ratio (EtTL), and exports to GDP ratio (EXtGDP) are the other important variables that increased the resiliency up to 6%, 5%, and 3%, respec- tively. Furthermore, for China, a 5% positive change in exports to GDP ratio (EXtGDP) has the best effect on LOMeR and increased resiliency up to 9%. Public debt to GDP ratio (PDtGDP), exchange rate (ER), and customer price index (CPI) is the other important variables that increased the resiliency up to 7%, 5%, and 4%, respectively. For Iran, a 5% positive change in exports to GDP ratio (EXtGDP) has the best effect on LOMeR and increased resiliency up to 7%. Customer price index (CPI) and exchange rate (ER) are the other important variables that increased resiliency by up to 6% and 5%, respectively.

5.2. Extending the data for the case of COVID-19

Due to the high importance of economic performance analysis in the recent COVID-19 shock, a data development was performed in this

Table 3 The results of socio-economic shock’s periods in the United States, China, and Iran.

Shock period United States China Iran

Start date End date Start date End date Start date End date

1 May 2007 September 2008 February 2008 August 2009 October 2009 January 2011 2 Nov 2010 February 2012 August 2011 January 2013 January 2014 August 2015 3 January 2013 February 2014 December 2013 September 2016 4 March 2014 October 2014 5 August 2015 March 2017

Table 4 The results of LOMeR for the three countries.

Time period LOMEFR Relevant socio-economic shock

United States Period 1 0.772 Global financial crisis Period 2 0.275 Inflation rate increasing Period 3 0.164 Oil price decreasing Period 4 0.126 Oil price decreasing Period 5 0.464 Debt to GDP ratio increasing

China Period 1 0.425 Global financial crisis Period 2 0.115 National Crackdown on corruption Period 3 0.493 National Stock market shock

Iran Period 1 0.244 Global financial crisis Period 2 0.364 Oil price decreasing

Table 5 The obtained variables’ weights for the United States.

Variable ri ci r-c (relation degree) r + c (prominence degree) Combined importance degree (SE) Rank of variable

PDtGDP 6.043 1.606 4.437 7.649 0.23154127 1 NPLtTL 3.242 1.43 1.812 4.672 0.11498557 5 EtTL 4.419 0.043 4.376 4.462 0.18769637 3 SMR 1.804 3.332 − 1.528 5.136 0.02278138 8 EXtGDP 1.921 2.012 − 0.091 3.933 0.04921286 6 CPI 2.101 4.039 − 1.938 6.14 0.02395812 7 PPI 0.942 2.128 − 1.186 3.07 0.0055908 9 UER 4.092 0.422 3.67 4.514 0.16759854 4 ER 0.969 2.384 − 1.415 3.353 0.00258317 10 ItR 5.044 1.292 3.752 6.336 0.19405192 2

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

10

section with the aim of analyzing the resilience of macroeconomics. Using data from 2019 to 2021, we observed that all three countries were affected by the shock. Since the developed model in calculating resil- ience (dynamically) requires a recovery time, the data and results show that the recovery period has not yet been completed in Iran. This course was completed first for China and then in the United States. Therefore, in calculating resilience according to Equation (16), it is necessary to explain that macroeconomic resilience in Iran has been done with the error of not completing the recovery period correctly. Figs. 9–11 show the MEP status of the three economies from 2019 to 2021, respectively. According to the diagrams, the values of performance reduction and recovery time (according to the proposed turning point algorithm) were calculated, and LOMER was determined. This amount of LOMER for

China, the United States, and Iran was 0.125, 0.315, and 0.298, respectively.

The importance of this analysis is highlighted when it is observed that according to the sensitivity analysis performed in the previous section, the most important factor affecting China’s resilience has been its export factor. A closer look at the export rate data showed that there was a 63% decrease in February 2020, which was compensated by a 148% increase after two months. Meanwhile, in the United States, where the most important factors have been the debt-to-GDP ratio and the unemployment rate, especially in terms of unemployment, there has been a sharp jump of about 12% in several consecutive months. Also, in Iran, the negative changes in the debt-to-GDP ratio and consumer prices in the recent periods have been significant, which makes sense to reduce

Table 6 The obtained variables’ weights for China.

Variable ri ci r-c (relation degree) r + c (prominence degree) Combined importance degree (SE) Rank of variable

PDtGDP 3.912 1.466 2.446 5.378 0.17371137 2 NPLtTL 3.252 2.543 0.709 5.795 0.1110711 6 EtTL 1.014 0.943 0.071 1.957 0.03085559 8 SMR 0.997 1.768 − 0.771 2.765 0.00917529 9 EXtGDP 4.912 1.439 3.473 6.351 0.22822794 1 CPI 4.033 3.461 0.572 7.494 0.13001192 4 PPI 3.432 2.212 1.22 5.644 0.12909296 5 UER 1.629 2.043 − 0.414 3.672 0.03627857 7 ER 2.726 0.137 2.589 2.863 0.14331136 3 ItR 1.197 2.232 − 1.035 3.429 0.0082639 10

Table 7 The obtained variables’ weights for Iran.

variable ri ci r-c (relation degree) r + c (prominence degree) Combined importance degree (SE) Rank of variable

PDtGDP 3.014 1.967 1.047 4.981 0.13564697 4 NPLtTL 1.823 2.926 − 1.103 4.749 0.01803477 8 EtTL 0.996 1.802 − 0.806 2.798 0.00235564 9 SMR 1.237 2.825 − 1.588 4.062 − 0.01871412 10 EXtGDP 2.097 0.208 1.889 2.305 0.13716224 3 CPI 5.329 1.025 4.304 6.354 0.33026213 1 PPI 2.18 1.299 0.881 3.479 0.10267319 6 UER 2.564 1.403 1.161 3.967 0.12536024 5 ER 3.247 2.313 0.934 5.56 0.13898339 2 ItR 1.004 1.115 − 0.111 2.119 0.02823554 7

Fig. 6. United States’ LOMeR improvement by 5% positive change in each variable.

Fig. 7. China’s LOMeR improvement by 5% positive change in each variable.

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

11

its resilience.

6. Research findings

The world economy has experienced various socio-economic shocks like global financial crisis, oil price change, terrorist attacks, trade war, national stock market shocks, and COVID-19 in recent years. The concept of resiliency has been used to assess macroeconomic conditions at various times and improve its conditions in recent years. One of the challenges of the studies presented in this field is the diversity in pre- senting the concept of resiliency and its computational approaches. Studies in the field of resiliency have typically explored regression models for countries at different times by identifying a set of indicators and developing regression models. The most important challenges of these papers have been the lack of attention to the nature of the iden- tified variables, the emphasis on qualitative data, the high diversity of resilience variables, the length of period, and the inefficiency of the measures.

Using Hellegate [4] and Pant et al. 0 [21]’s original idea in devel- oping a graph for resiliency analysis, this study takes the following steps to measure macroeconomic resilience.

1 Identifying effective variables that affect macroeconomic resilience; 2-Integrating variables to develop macroeconomic performance in- dicator from a resilience perspective;

3 Determining the periods of shocks according to the variation of macroeconomic performance by using an algorithm to detect turning points;

4 Measuring macroeconomic resiliency in shock periods; 5 Analyzing the affective variables and reviewing strategies to improve

resiliency.

Since the nature of macroeconomic resilience and macroeconomic performance should be quantitative, in the first part, quantitative vari- ables with shorter reporting periods were attempted to provide macro- economic performance information at appropriate time intervals. Based on the available data, we use monthly data for macroeconomic perfor- mance calculation.

In the second part, these variables are integrated using data envel- opment analysis to determine the macroeconomic performance measure for each period. The third section also identifies the macroeconomic performance diagrams. Previous resiliency studies have defined resil- iency periods as being considered from the onset of the crisis (i.e., the start of a decline in performance) until the full returns to its previous level of crisis. Since, in many cases, it has been seen that the value of performance after the crisis has not returned to its beginning level, here to choose the shocks’ periods, we develop an algorithm to detect the turning points. Our proposed algorithm considers both a jump in observation and a sequence of consecutive negative observations. Like that, if the Macroeconomic Performance (MEP) indicator changes in a month have a large value, it is likely that the shock period will begin from that time. In addition, the shock period begins if there are consistent decreasing observations. Similarly, the consecutive positive observations may lead to the end of the shock. The next step is to develop a function to accommodate the amount of macroeconomic resilience. This function calculates the total amount of resiliency loss in these periods based on the graph of the resilience function and the calculation area of the subsurface of the shock periods. To this point, the amount of resilience of the macroeconomic in critical times has been determined, but the issue of upgrading it should also be emphasized. To this end, by analyzing the relationships and impacts of the effective variables on each other, more important variables have been identified and examined to what extent a positive change of 5% affects resiliency. To show the applicability of our proposed approach, data from three countries of the US, China and Iran were used, and the macroeconomic resilience of them due to the most important recent socio-economic shocks, including the COVID-19 pandemic, was also examined. Analyzing the resiliency and the used variables revealed that for the US, a 5% positive change in Public debt to GDP ratio, Unemployment rate, equity to total asset ratio, and Exports to GDP ratio has the best effects on Loss of Macroeconomic Resiliency (LOMeR). Furthermore, for China, a 5% positive change in Exports to GDP ratio, Public debt to GDP ratio, Exchange rate, and Consumer price index have the best effect on LOMeR. For Iran, a 5% positive change in Export to GDP ratio, Consumer price index, and Exchange rate have the best effects on LOMeR.

The most important advantage of our proposed approach is the ability to present a resiliency status at any given time-sensitive to changes in variables. Furthermore, by examining the relationships be- tween variables concerning the CoVaR measure and the DEMATEL method, it is possible to help policymakers decide to improve macro- economic resiliency by focusing more on the important variables.

Future studies in this area can explore remedial measures using other approaches and examining the type of crisis. These studies could also provide other approaches to identifying crisis periods. One of the limi- tations of this paper has been the need to use quantitative data. Other studies can use qualitative data in addition to quantitative data by appropriate approaches.

Author Statement

Hojat Rezaei Soufi: Conceptualization, Data curation, Investigation, Methodology, Software, Visualization, Writing – original draft. Akbar Esfahanipour: Conceptualization, Methodology, Supervision, Valida- tion, Writing – review & editing. Mohsen Akbarpour Shirazi: Concep- tualization, Methodology, Validation

Fig. 8. Iran’s LOMeR improvement by 5% positive change in each variable.

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

12

Table 8 Results of LOMeR with a 5% positive change in variables.

variable USA China Iran

Results of LOMeR with 5% positive change on average of relevant indicator

Results of LOMeR with 5% positive change on average of relevant indicator

Results of LOMeR with 5% positive change on average of relevant indicator

PDtGDP Period 1 0.697 Period 1 0.400 Period 1 0.243 Period 2 0.250 Period 3 0.154 Period 2 0.106 Period 2 0.343 Period 4 0.116 Period 5 0.422 Period 3 0.456 The average percent of improvement 3.13% The average percent of improvement 8.47% The average percent of improvement 7.09%

NPLtTL Period 1 0.744 Period 1 0.423 Period 1 0.242 Period 2 0.265 Period 3 0.163 Period 2 0.111 Period 2 0.359 Period 4 0.122 Period 5 0.462 Period 3 0.471 The average percent of improvement 1.11% The average percent of improvement 2.34% The average percent of improvement 2.82%

EtTL Period 1 0.716 Period 1 0.420 Period 1 0.241 Period 2 0.260 Period 3 0.154 Period 2 0.112 Period 2 0.362 Period 4 0.122 Period 5 0.443 Period 3 0.486 The average percent of improvement 0.78% The average percent of improvement 5.34% The average percent of improvement 1.65%

SMR Period 1 0.763 Period 1 0.423 Period 1 0.242 Period 2 0.280 Period 3 0.163 Period 2 0.114 Period 2 0.363 Period 4 0.124 Period 5 0.462 Period 3 0.486 The average percent of improvement 0.44% The average percent of improvement 0.43% The average percent of improvement 1.07%

EXtGDP Period 1 0.744 Period 1 0.375 Period 1 0.226 Period 2 0.265 Period 3 0.159 Period 2 0.103 Period 2 0.337 Period 4 0.122 Period 5 0.443 Period 3 0.456 The average percent of improvement 7.38% The average percent of improvement 3.64% The average percent of improvement 9.81%

CPI Period 1 0.744 Period 1 0.400 Period 1 0.236 Period 2 0.270 Period 3 0.162 Period 2 0.110 Period 2 0.332 Period 4 0.122 Period 5 0.443 Period 3 0.478 The average percent of improvement 6.08% The average percent of improvement 2.97% The average percent of improvement 4.48%

PPI Period 1 0.763 Period 1 0.418 Period 1 0.242 Period 2 0.270 Period 3 0.163 Period 2 0.111 Period 2 0.348 Period 4 0.124 Period 5 0.462 Period 3 0.478 The average percent of improvement 2.58% The average percent of improvement 1.15% The average percent of improvement 2.72%

UER Period 1 0.706 Period 1 0.418 Period 1 0.243 Period 2 0.260 Period 3 0.154 Period 2 0.112 Period 2 0.359 Period 4 0.120 Period 5 0.422 Period 3 0.486 The average percent of improvement 0.92% The average percent of improvement 6.83% The average percent of improvement 1.84%

ER Period 1 0.763 Period 1 0.400 Period 1 0.235 Period 2 0.275 Period 3 0.163 Period 2 0.110 Period 2 0.337 Period 4 0.124 Period 5 0.460 Period 3 0.464 The average percent of improvement 5.53% The average percent of improvement 0.92% The average percent of improvement 5.46%

ItR Period 1 0.746 Period 1 0.420 Period 1 0.240 Period 2 0.266 Period 3 0.158 Period 2 0.114 Period 2 0.359 Period 4 0.122 Period 5 0.451 Period 3 0.492 The average percent of improvement 1.48% The average percent of improvement 3.25% The average percent of improvement 0.87%

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

13

Fig. 9. Macroeconomic Performance (MEP) for the US from 2019 to 2021.

Fig. 10. Macroeconomic Performance (MEP) for China from 2019 to 2021.

Fig. 11. Macroeconomic Performance (MEP) for Iran from 2019 to 2021.

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

14

Appendix A

The results of MEP calculations for the three countries are presented in tables A.1, A.2, and A.3 for the United States, China, and Iran.

Table A.1 MEP results for United States

Date MEP Date MEP Date MEP Date MEP

06-Jan 0.75 09-Jan 0.17 12-Jan 0.48 15-Jan 0.88 06-Feb 0.72 09-Feb 0.19 12-Feb 0.57 15-Feb 0.87 06-Mar 0.77 09-Mar 0.28 12-Mar 0.59 15-Mar 0.94 06-Apr 0.79 09-Apr 0.26 12-Apr 0.55 15-Apr 0.98 06-May 0.8 09-May 0.27 12-May 0.52 15-May 1.03 06-Jun 0.82 09-Jun 0.38 12-Jun 0.49 15-Jun 0.95 06-Jul 0.88 09-Jul 0.34 12-Jul 0.58 15-Jul 0.89 06-Aug 0.85 09-Aug 0.41 12-Aug 0.62 15-Aug 0.82 06-Sep 0.89 09-Sep 0.44 12-Sep 0.69 15-Sep 0.79 06-Oct 0.9 09-Oct 0.52 12-Oct 0.78 15-Oct 0.75 06-Nov 0.92 09-Nov 0.57 12-Nov 0.84 15-Nov 0.76 06-Dec 0.87 09-Dec 0.53 12-Dec 0.82 15-Dec 0.64 07-Jan 0.83 10-Jan 0.58 13-Jan 0.75 16-Jan 0.62 07-Feb 0.8 10-Feb 0.67 13-Feb 0.74 16-Feb 0.58 07-Mar 0.81 10-Mar 0.66 13-Mar 0.72 16-Mar 0.59 07-Apr 0.75 10-Apr 0.74 13-Apr 0.64 16-Apr 0.55 07-May 0.69 10-May 0.76 13-May 0.61 16-May 0.62 07-Jun 0.62 10-Jun 0.83 13-Jun 0.58 16-Jun 0.66 07-Jul 0.54 10-Jul 0.85 13-Jul 0.54 16-Jul 0.66 07-Aug 0.48 10-Aug 0.89 13-Aug 0.55 16-Aug 0.64 07-Sep 0.37 10-Sep 0.84 13-Sep 0.59 16-Sep 0.67 07-Oct 0.28 10-Oct 0.72 13-Oct 0.64 16-Oct 0.62 07-Nov 0.19 10-Nov 0.7 13-Nov 0.54 16-Nov 0.59 07-Dec 0.12 10-Dec 0.75 13-Dec 0.58 16-Dec 0.66 08-Jan 0.05 11-Jan 0.73 14-Jan 0.63 17-Jan 0.71 08-Feb 0.07 11-Feb 0.69 14-Feb 0.69 17-Feb 0.73 08-Mar 0.09 11-Mar 0.65 14-Mar 0.74 17-Mar 0.75 08-Apr 0.08 11-Apr 0.68 14-Apr 0.64 17-Apr 0.71 08-May 0.14 11-May 0.62 14-May 0.6 17-May 0.68 08-Jun 0.16 11-Jun 0.57 14-Jun 0.52 17-Jun 0.67 08-Jul 0.15 11-Jul 0.61 14-Jul 0.47 17-Jul 0.69 08-Aug 0.19 11-Aug 0.55 14-Aug 0.45 17-Aug 0.72 08-Sep 0.21 11-Sep 0.52 14-Sep 0.51 17-Sep 0.75 08-Oct 0.24 11-Oct 0.48 14-Oct 0.59 17-Oct 0.71 08-Nov 0.26 11-Nov 0.42 14-Nov 0.67 17-Nov 0.74 08-Dec 0.21 11-Dec 0.41 14-Dec 0.79 17-Dec 0.7

Table A.2 MEP results for China

Date MEP Date MEP Date MEP Date MEP

06-Jan 0.73 09-Jan 0.43 12-Jan 0.52 15-Jan 0.45 06-Feb 0.73 09-Feb 0.39 12-Feb 0.5 15-Feb 0.48 06-Mar 0.74 09-Mar 0.34 12-Mar 0.48 15-Mar 0.46 06-Apr 0.73 09-Apr 0.37 12-Apr 0.47 15-Apr 0.44 06-May 0.75 09-May 0.39 12-May 0.45 15-May 0.43 06-Jun 0.77 09-Jun 0.38 12-Jun 0.46 15-Jun 0.38 06-Jul 0.78 09-Jul 0.38 12-Jul 0.46 15-Jul 0.39 06-Aug 0.81 09-Aug 0.37 12-Aug 0.45 15-Aug 0.37 06-Sep 0.8 09-Sep 0.39 12-Sep 0.47 15-Sep 0.38 06-Oct 0.8 09-Oct 0.41 12-Oct 0.47 15-Oct 0.36 06-Nov 0.79 09-Nov 0.42 12-Nov 0.48 15-Nov 0.35 06-Dec 0.83 09-Dec 0.44 12-Dec 0.49 15-Dec 0.33 07-Jan 0.82 10-Jan 0.46 13-Jan 0.48 16-Jan 0.33 07-Feb 0.81 10-Feb 0.47 13-Feb 0.5 16-Feb 0.32 07-Mar 0.79 10-Mar 0.48 13-Mar 0.5 16-Mar 0.31 07-Apr 0.82 10-Apr 0.51 13-Apr 0.81 16-Apr 0.33 07-May 0.84 10-May 0.52 13-May 0.79 16-May 0.31 07-Jun 0.87 10-Jun 0.52 13-Jun 0.77 16-Jun 0.3 07-Jul 0.89 10-Jul 0.54 13-Jul 0.75 16-Jul 0.33 07-Aug 0.88 10-Aug 0.55 13-Aug 0.73 16-Aug 0.34 07-Sep 0.86 10-Sep 0.54 13-Sep 0.74 16-Sep 0.35 07-Oct 0.86 10-Oct 0.57 13-Oct 0.69 16-Oct 0.35 07-Nov 0.87 10-Nov 0.58 13-Nov 0.65 16-Nov 0.36 07-Dec 0.86 10-Dec 0.55 13-Dec 0.66 16-Dec 0.35 08-Jan 0.82 11-Jan 0.56 14-Jan 0.47 17-Jan 0.35

(continued on next page)

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

15

Table A.2 (continued )

Date MEP Date MEP Date MEP Date MEP

08-Feb 0.82 11-Feb 0.56 14-Feb 0.48 17-Feb 0.37 08-Mar 0.78 11-Mar 0.57 14-Mar 0.46 17-Mar 0.38 08-Apr 0.76 11-Apr 0.6 14-Apr 0.46 17-Apr 0.4 08-May 0.74 11-May 0.61 14-May 0.45 17-May 0.42 08-Jun 0.71 11-Jun 0.6 14-Jun 0.44 17-Jun 0.41 08-Jul 0.67 11-Jul 0.58 14-Jul 0.47 17-Jul 0.39 08-Aug 0.64 11-Aug 0.55 14-Aug 0.44 17-Aug 0.42 08-Sep 0.59 11-Sep 0.55 14-Sep 0.43 17-Sep 0.43 08-Oct 0.57 11-Oct 0.53 14-Oct 0.42 17-Oct 0.42 08-Nov 0.52 11-Nov 0.53 14-Nov 0.44 17-Nov 0.44 08-Dec 0.49 11-Dec 0.53 14-Dec 0.44 17-Dec 0.45

Table A.3 MEP results for Iran

Date MEP Date MEP Date MEP Date MEP

06-Jan 0.66 09-Jan 0.7 12-Jan 0.62 15-Jan 0.44 06-Feb 0.66 09-Feb 0.71 12-Feb 0.66 15-Feb 0.39 06-Mar 0.68 09-Mar 0.73 12-Mar 0.67 15-Mar 0.37 06-Apr 0.71 09-Apr 0.72 12-Apr 0.69 15-Apr 0.35 06-May 0.73 09-May 0.74 12-May 0.71 15-May 0.41 06-Jun 0.75 09-Jun 0.75 12-Jun 0.72 15-Jun 0.43 06-Jul 0.77 09-Jul 0.76 12-Jul 0.73 15-Jul 0.45 06-Aug 0.76 09-Aug 0.78 12-Aug 0.75 15-Aug 0.47 06-Sep 0.75 09-Sep 0.77 12-Sep 0.74 15-Sep 0.48 06-Oct 0.79 09-Oct 0.79 12-Oct 0.77 15-Oct 0.51 06-Nov 0.81 09-Nov 0.81 12-Nov 0.77 15-Nov 0.55 06-Dec 0.77 09-Dec 0.8 12-Dec 0.79 15-Dec 0.55 07-Jan 0.78 10-Jan 0.58 13-Jan 0.77 16-Jan 0.54 07-Feb 0.77 10-Feb 0.57 13-Feb 0.79 16-Feb 0.53 07-Mar 0.77 10-Mar 0.55 13-Mar 0.81 16-Mar 0.56 07-Apr 0.76 10-Apr 0.53 13-Apr 0.8 16-Apr 0.58 07-May 0.74 10-May 0.54 13-May 0.77 16-May 0.62 07-Jun 0.74 10-Jun 0.55 13-Jun 0.76 16-Jun 0.64 07-Jul 0.75 10-Jul 0.53 13-Jul 0.72 16-Jul 0.66 07-Aug 0.76 10-Aug 0.5 13-Aug 0.74 16-Aug 0.64 07-Sep 0.77 10-Sep 0.5 13-Sep 0.75 16-Sep 0.67 07-Oct 0.77 10-Oct 0.48 13-Oct 0.77 16-Oct 0.68 07-Nov 0.78 10-Nov 0.47 13-Nov 0.72 16-Nov 0.69 07-Dec 0.79 10-Dec 0.49 13-Dec 0.74 16-Dec 0.66 08-Jan 0.77 11-Jan 0.49 14-Jan 0.75 17-Jan 0.71 08-Feb 0.78 11-Feb 0.52 14-Feb 0.73 17-Feb 0.72 08-Mar 0.72 11-Mar 0.54 14-Mar 0.7 17-Mar 0.74 08-Apr 0.74 11-Apr 0.57 14-Apr 0.66 17-Apr 0.7 08-May 0.73 11-May 0.57 14-May 0.67 17-May 0.69 08-Jun 0.72 11-Jun 0.59 14-Jun 0.65 17-Jun 0.66 08-Jul 0.7 11-Jul 0.61 14-Jul 0.62 17-Jul 0.65 08-Aug 0.71 11-Aug 0.63 14-Aug 0.61 17-Aug 0.64 08-Sep 0.68 11-Sep 0.65 14-Sep 0.57 17-Sep 0.66 08-Oct 0.69 11-Oct 0.66 14-Oct 0.52 17-Oct 0.68 08-Nov 0.69 11-Nov 0.64 14-Nov 0.47 17-Nov 0.69 08-Dec 0.67 11-Dec 0.64 14-Dec 0.46 17-Dec 0.71

References

[1] Briguglio L, Cordina G, Farrugia N, Vella S. Economic vulnerability and resilience: concepts and measurements. Oxf Dev Stud 2008;37(3):1810–2611.

[2] Buheji M. Understanding the Power of Resilience Economy: An Inter-Disciplinary Perspective to Change the World Attitude to Socio-Economic Crisis. AuthorHouse; 2018.

[3] Cabezon E, Hunter L, Tumbarello P, Washimi K, Wu Y. Enhancing macroeconomic resilience to natural disasters and climate change in the small states of the Pacific. Asian Pac Econ Lit 2019;33(1):113–30.

[4] Hallegatte S. Economic resilience: definition and measurement, World Bank policy research working paper No. 6852. https://papers.ssrn.com/sol3/papers.cfm?abs tract_id=2432352; 2014.

[5] Allen F, Gu X, Kowalewski O. Financial crisis, structure and reform. Wharton School WP Series, University of Pennsylvania; 2011.

[6] Soufi HR, Esfahanipour A. Modeling financial resilience in commercial banks using Multinomial Logistic regression. 17th Iranian international industrial engineering conference. 2021.

[7] Jethwaney J. Covid-19 disaster: interdependence of crisis communication and socio-economic resilience. In: Disaster management for 2030 Agenda of the SDG. Singapore: Palgrave Macmillan; 2020. p. 333–58.

[8] Okuyama Y, Hewings GJ, Sonis M. Measuring economic impacts of disasters: interregional input-output analysis using sequential interindustry model. Model. Spat. Econ. Impacts Disasters 2004:77–101.

[9] Rose A. Economic resilience to disasters (CARRI research report 8). Community and Regional Resilience Institute; 2009.

[10] Billio M, Getmansky M, Lo AW, Pelizzon L. Econometric measures of connectedness and systemic risk in the finance and insurance sectors. J Financ Econ 2012;104:535–59.

[11] Schätter F, Hansen O, Wiens M, Schultmann F. A decision support methodology for a disaster-caused business continuity management. Decis Support Syst 2019;118: 10–20.

[12] Kamissoko D, Benaben F, Amendeep A, Nastov B, Chapurlat V, Bony-Dandrieuxd A, Daclin N. May). A Decision Support System for resilience based on functionality analysis of interconnected systems. In: EmC-ICDSST 2019 5 th international conference on decision support system Technology–ICDSST 2019 and EURO Mini conference 2019 on “decision support systems: main; 2019.

H. Rezaei Soufi et al.

Socio-Economic Planning Sciences 79 (2022) 101101

16

[13] Briguglio L, Cordina G, Farrugia N, Vella S. Conceptualising and measuring economic reslience. In: Briguglio L, Cordina G, Kisanga EJ, editors. Building the economic Reslience of small states, Malta: Islands and small states. Institute of the University of Malta and London: Commonwealth Secretariat; 2006. p. 265–88.

[14] Boorman J, Fajgenbaum J, Ferhani H, Bhaskaran M, Arnold D, Kohli HA. The Centennial resilience index: measuring countries’ resilience to shock. Global Journal of Emerging Market Economies 2013;5(2):57–98.

[15] Boorman J. The IMF and the crises in Greece, Ireland, and Portugal: an evaluation by the independent evaluation Office—Summary of views of the Advisory group. IEO background paper No. BP/16-02/01 (Washington: international Monetary Fund). 2016.

[16] Angulo AM, Mur J, Trívez FJ. Measuring resilience to economic shocks: an application to Spain. Ann Reg Sci 2018;60(2):349–73.

[17] Röhn O, Sánchez AC, Hermansen M, Rasmussen M. Economic resilience: a new set of vulnerability indicators for OECD countries. 2015.

[18] Mirzaei A, Al-Khouri RSF. The resilience of oil-rich economies to the global financial crisis: evidence from Kuwaiti financial and real sectors. Econ Syst 2016; 40(1):93–108.

[19] Marto R, Papageorgiou C, Klyuev V. Building resilience to natural disasters: an application to small developing states. J Dev Econ 2018;135:574–86.

[20] Rose A. Defining and measuring economic resilience to disasters. Disaster Prev Manag: Int J 2004;13(4).

[21] Pant R, Barker K, Zobel CW. Static and dynamic metrics of economic resilience for interdependent infrastructure and industry sectors. Reliab Eng Syst Saf 2014;125: 92–102.

[22] Soufi HR, Esfahanipour A, Shirazi MA. A quantitative measure for financial resilience of firms: Evidence from Tehran stock exchange. Scientia Iranica; 2021. https://doi.org/10.24200/sci.2021.55845.4433.

[23] Team AWM Strategy. Community economic resilience index. 2010. Birmingham, UK.

[24] Sahoo BK, Acharya D. Constructing macroeconomic performance index of Indian states using DEA. J Econ Stud 2012;39(1):63–83.

[25] Mohanty RK, Sahoo BK. Examining the Eco-macroeconomic performance index of India: a data envelopment analysis approach (No. 17/202). 2017.

[26] Pelling M, Özerdem A, Barakat S. The macro-economic impact of disasters. Prog Dev Stud 2002;2:283–305.

[27] Cissé JD, Barrett CB. Estimating development resilience: a conditional moments- based approach. J Dev Econ 2018;135:272–84.

[28] Peykani P, Mohammadi E, Saen RF, Sadjadi SJ, Rostamy-Malkhalifeh M. Data envelopment analysis and robust optimization: a review. Expet Syst 2020. https:// doi.org/10.1111/exsy.12534.

[29] Zobel CW, Baghersad M. Analytically comparing disaster resilience across multiple dimensions. Soc Econ Plann Sci 2020;69:100678.

[30] Lapuh L. Socio-economic characteristics of resilient localities–experiences from Slovenia. Regional Studies, Regional Science 2018;5(1):149–56.

[31] Yin J, Si YW, Gong Z. June). Financial time series segmentation based on Turning Points. In: Proceedings 2011 international conference on system science and engineering. IEEE; 2011. p. 394–9.

[32] Ocampo L, Yamagishi K. Modeling the lockdown relaxation protocols of the Philippine government in response to the COVID-19 pandemic: an intuitionistic fuzzy DEMATEL analysis. Soc Econ Plann Sci 2020:100911.

[33] Chang B, Chang CW, Wu CH. Fuzzy DEMATEL method for developing supplier selection criteria. Expert Syst Appl 2011;38(3):1850–8.

[34] Adrian T, Brunnermeier MK. CoVaR NBER working paper series. 2011. p. w17454. [35] Reboredo JC, Ugolini A. Systemic risk in European sovereign debt markets: a

CoVaR-copula approach. J Int Money Finance 2015;51:214–44. [36] Mangla S, Kumar P, Barua MK. An evaluation of attribute for improving the green

supply chain performance via DEMATEL method. International Journal of Mechanical Engineering and Robotics Research 2014;1(1):30–5.

Hojat rezaei Soufi received his B.Sc in Industrial Engineering in 2012 from Khaje Nasire- Tousi University and M.Sc in industrial engineering in 2015 from Tehran University. He is currently PhD Candidate at Amirkabir University of Technology, Tehran, Iran. He has worked on Risk management, Business Continuity Management, Financial Resilience, and decision making methods during 2008–2020 and has published 16 scientific articles in prestigious international journals and conferences proceedings.

Akbar Esfahanipour received his B.Sc in Industrial engineering from Amirkabir Univer- sity of Technology, Tehran, Iran in 1995. His M.Sc and PhD degrees are in Industrial Engineering from Tarbiat Modares University, Tehran, Iran in 1998 and 2004, respec- tively. After that, he was a Postdoctoral Fellow in the field of Management Information Systems at DeGroote School of Business, McMaster University, Hamilton, ON, Canada. He is currently an Associate Professor at Department of Industrial Engineering and Manage- ment Systems, Amirkabir University of Technology. His teaching and research activities are mainly in the fields of Financial Engineering, Risk Analysis and Artificial Intelligence. His research interests are in the areas of forecasting in financial markets, application of soft computing methods in financial decision making, behavioral finance, and analysis of financial risks. Dr. Esfahanipour has published more than 70 research articles in presti- gious academic journals as well as in conference proceedings.

Mohsen Akbarpour Shirazi received his B.Sc in Industrial engineering from Isfahan University of Technology, Isfahan, Iran in 1990. His M.Sc and PhD degrees are in Industrial Engineering from Amirkabir University of Technology, Tehran, Iran in, 1993 and 2001, respectively. He is currently an Associate Professor at Department of Industrial Engi- neering and Management Systems, Amirkabir University of Technology, Tehran, Iran. His areas of research include supply chain planning, transportation, and systems modeling. He is the author and co-author of many research papers in these fields.

H. Rezaei Soufi et al.