Comovements in financial fluctuations are anchored to economic fundamentals: A mesoscopic mapping
Abstract
We demonstrate the existence of an empirical linkage between the nominal financial networks and the underlying economic fundamentals across countries. We construct the nominal return correlation networks from daily data to encapsulate sectorlevel dynamics and figure the relative importance of the sectors in the nominal network through a measure of centrality and clustering algorithms. The eigenvector centrality robustly identifies the backbone of the minimum spanning tree defined on the return networks as well as the primary cluster in the multidimensional scaling map. We show that the sectors that are relatively large in size, defined with the metrics market capitalization, revenue and number of employees, constitute the core of the return networks, whereas the periphery is mostly populated by relatively smaller sectors. Therefore, sectorlevel nominal return dynamics is anchored to the real size effect, which ultimately shapes the optimal portfolios for risk management. Our results are reasonably robust across 27 countries of varying degrees of prosperity and across periods of market turbulence (200809) as well as relative calmness (201516).
1 Introduction
Widespread existence of bubbles in the financial markets and extreme movements of return series indicate that the relationship between the macroeconomic fundamentals and the asset prices is unstable [1]. The ‘excess volatility puzzle’ in the stock markets refers precisely to this disconnect between the volatility of asset returns and the movements of the underlying fundamentals [2]. Recent research emphasizes the roles played by wrong expectation, bounded rationality, herding behavior, etc. as being important causal factors for the observed disconnect [3]. In this paper, we present an alternate view that the comovements in financial assets are anchored to the corresponding macroeconomic fundamentals. Thus, nominal returns from individual assets might drift far from what can be predicted using expected cashflow, while the joint evolution of the comovement of returns are still related to aggregate size variables like market capitalization, revenue or number of employees.
In the following, we consider the economy to be a multilayered network [4] defined over nodes at different levels of granularity, each having significantly different properties. At the micro level, firm size distributions show power law decays [5] and biexponential growth size distributions [6]. A scaling relationship between size of the firms and the corresponding volatiltity was also proposed [6]. At the macro level, similar features are seen, for example, as has been proposed by [7]. These suggest that there might be universal features of growth processes of economic entities (see also [8]). Ref. [7] also argued that the dispersion in relative sizes of firms contribute substantially to the aggregate volatility of an economy, providing a link from the micro level to the macro level. A complementary view has emerged from the network literature that the dynamics at the intermediate sectoral level could play an important role in shaping the aggregate macrolevel dynamics [9]. We focus precisely on the ‘mesoscopic’ level, which identifies with the production process of the economy while being granular enough to capture the network structure of comovements in return fluctuations.
There are two modes of connectedness across sectors. At the nominal pricing level, the fluctuations of returns from the sectoral indices show the degree of comovements across sectors. At the production level, the flow of goods and services across sectors [10] gives rise to dispersion in relative sizes of these sectors. Here, we show that there exists a universal mapping between the intersectoral return dynamics and relative sizes of the sectors defined with multiple metrics, thus highlighting an empirical link between financial networks and macroscopic variables in a granular economy. In particular, we show that the sectors with disproportionate shares of the economy, constitute the core of the corresponding return networks. Therefore, at the ‘mesoscopic’ level, the dispersion in size explains the dispersion in ‘centralities’ of nominal fluctuations of sectors.
To study the topology of the return network, we construct return correlation matrices from sectoral indices for 27 countries, and apply two commonly used clustering algorithms (minimum spanning tree and multidimensional scaling) to group sectors based on their comovements. The influence of the sectors in the whole network can be found by using the eigenvector centrality, which is able to handle both directed as well as weighted graphs [11]. In this paper, we also propose a methodology to find a binary characterization of the ‘coreperiphery’ structure by using a modification of the eigenvector centrality. Such classification of the sectors according to whether they belong to the core or the periphery, allows one to pin down exactly which sectors are driving the market correlations. We show that these sectors identified as core by the centrality measure, also constitute the backbone of the minimum spanning tree (MST) and cluster very closely in multidimensional scaling (MDS) maps, thereby confirming the robustness of our method of extraction of the coreperiphery structure.
To study the connection between the financial network with the underlying production process, we regress the eigenvector centrality measure on sector sizes defined with three different metrics, viz., market capitalization, revenue and employment, all aggregated at the sectoral level. The results across 27 countries clearly indicate that the dispersion in the economic size explains the variation in the dispersion of sectoral centralities in the correlation matrix. This is the primary finding of our paper, as it establishes the the linkage between the economic fundamentals and the fluctuations of the return series. Finally, we study the risk diversification of a portfolio comprising sectoral indices, based on the eigenvector centralities. For the sake of simplicity, we use a rudimentary Markowitz portfolio allocation problem and show that the core sectors, i.e., the ones with sufficiently high centralities, do not usually appear in a minimum variance portfolio. Intuitively, very large sectors contribute significantly to the movement of the return correlations and they constitute the ‘market factor’ of correlations. Hence, for reduction of the volatility of the portfolio, the weights assigned to such sectors contributing to the aggregate risk, are necessarily minimized.
We perform statistical tests on a comprehensive list of 27 countries that includes developed as well as developing countries across five continents, totaling 72 sectors in the financial economies. We base most of our studies on a recent and relatively calm period (201516), and then compare and contrast with a volatile period (200809), in order to check robustness of our findings across time. We show that the 201516 period gives very consistent results (25 out of 27 countries are in expected direction), whereas 200809 period is largely consistent (22 out of 26), although there are some aberrations as the number of statistically insignificant relationships increases. A consistent pathogenic case is Greece, which has been known to possess weak economic fundamentals along with severe crises in the financial markets in the recent times.
2 Data, Definitions and Methods
2.1 Data Description
We have used the sectoral price indices from the Thomson Reuters Eikon database [12], within the time frames January 2008 December 2009, and October 2014 September 2016. We have analyzed the data for a total of 72 sectors (see table 1), for the following countries: (1) AUS Australia (2) BEL Belgium (3) CAN Canada (4) CHE Switzerland (5) DEU Germany (6) DNK Denmark (7) ESP Spain (8) FIN Finland (9) FRA France (10) GBR United Kingdom (11) GRC Greece (12) HKG Hong Kong (13) IDN Indonesia (14) IND India (15) JPN Japan (16) LKA Sri Lanka (17) MYS Malaysia (18) NLD the Netherlands (19) NOR Norway (20) PHL Philippines (21) PRT Portugal (22) QAT Qatar (23) SAU Saudi Arabia (24) SWE Sweden (25) THA Thailand (26) USA United States of America and (27) ZAF South Africa, spread across the continents of the Americas, Europe, Africa, Asia and Australia. The time series data on the real variables, such as market capitalization, revenue and the number of employees within each sector, are also available in the same database although at the company level rather than at the sectoral level. Hence, for our purposes of constructing sectorlevel macro aggregate variables, we collected the companies listed within each sector for one particular country, and then aggregated the relevant companyspecific variables across all such companies within the corresponding sector.
We find that the USA economy is a good representative of the empirical results and hence, in the main text, we present the results for the USA economy in details. For the other 26 countries, the detailed results are presented in the Supplementary material. Note that data for Finland (FIN) was not available for the period 200809.
Label  Sector  Label  Sector 

AF  Agro & Food Industry  MD  Media 
AG  Agriculture  MF  Manufacturing 
AM  Automobiles  MG  Mining 
BC  Building & Construction  MI  Multi Investments 
BF  Banks & Finance  MID  Miscellaneous Industries 
BFT  Beverage, Food & Tobacco  MM  Metals & Mining 
BK  Bank  MO  Mining & Oil 
BM  Basic Materials  MOT  Motors 
BR  Basic Resources  MP  Metal Products 
CC  Consumer & Cyclical  MP1  Media & Publishing 
CD  Consumer Discretionary  MT  Media & Telecomm 
CD1  Consumer Durables  OC  Oil & Coal Products 
CE  Cement  OG  Oil and Gas 
CG  Consumer Goods  PC  Property & Construction 
CG1  Capital Goods  PE  Power & Energy 
IT  Information Technology  PG  Personal Goods 
CH  Chemicals  PH  PetroChemicals 
CM  Consturction & Materials  PL  Plantation 
CN  Construction  PR  Property 
CP  Consumer Products  PSU  Public Sector Undertaking 
CS  Consumer Staples  RB  Rubber 
CSR  Consumer Services  RE  Real Estate 
EC  Energy & Chemical  RT  Retail 
EG  Energy  RY  Realty 
EM  Electrical Machinery  SC  Semiconductor 
EU  Energy & Utilities  ST  Steel 
FB  Food & Beverages  SU  Securities 
FN  Finance  TC  Telecom 
GD  Gold  TD  Trade 
HC  Health Care  TE  Transport & Equipment 
HG  Household Goods  TP  Transport 
HT  Hotel & Tourism  TS  Trade & Services 
ID  Industries  TT  Travel & Tourism 
IF  Infrastructure  TX  Textiles 
IP  Industrial Production  UT  Utilities 
IS  Insurance  WS  Wholesale 
2.2 Correlation coefficient and the distance metric
If represents the return of sectors, which is calculated as , where is the adjusted closure price of sector in day , then the equal time Pearson correlation coefficients between sectors and is defined as
(1) 
where represents the expectation. We use to denote the return correlation matrix.
2.3 Eigenvector centrality
To analyze the influence of a sector in the whole network, the ranking of the sectors is measured by the eigenvector centrality. It is not necessary that a sector with high eigenvector centrality is highly linked but the sector might have few but important links. Given an matrix , the eigenvector centrality is defined as an vector , which solves
(2) 
where is the dominant eigenvalue of .
In general, almost all pairwise correlations are positive. However, in rare cases (e.g., Gold sector in Canada), certain sectors show mild negative correlations with other sectors. We consider the absolute value of the correlation matrix for computing the eigenvector centrality, since according to the PerronFrobenius theorem, a real square matrix with positive entries has a unique largest real eigenvalue and the corresponding eigenvector has strictly positive components. Finally, we normalize the centrality vector such that = 1.
We consider a further modification of the centrality measure to identify the coreperiphery structure in a binary fashion. Instead of the level values of the correlation coefficients, we consider , where is a sufficiently large even number, since this transformation would make the many weak correlations have asymptotically zero weights while maintaining positive signs. We found that is the lowest value, which gives reasonably good estimates of the backbone of the minimum spanning tree. Hence, we present results for although, in principle, one can use higher values as well. To determine the core sectors of a country, we then construct a threshold value , as a fixed percentage of the coefficient of variation (standard deviation/mean) for the eigenvector centralities. If the sectoral centrality is above the threshold value , then the sector is considered as core, otherwise not.
2.4 Multidimensional scaling
To analyze the similarity among different sectors in terms of distances (), geometrical maps are generated using MDS for each of 27 countries, where each sector corresponds to a set of coordinates in a multidimensional space. The concept behind MDS is to represent two similar sectors as two sets of coordinates that are close to each other, and two sectors behaving differently are placed far apart in the space [16]. Given , the aim of MDS is to generate vectors , such that
(3) 
where represents vector norm. To plot the vectors in the form of a map, the embedding dimension is chosen as . Generally, MDS can be obtained through an optimization problem, where () is the solution of the problem of minimization of a cost function, such as
(4) 
2.5 Minimum spanning tree (MST)
MST is a clustering algorithm. By giving distance matrix as input, MSTs are constructed for sectors for each of the 27 countries, which are connected, undirected graphs such that all the sectors are connected together with the minimal total weighting for its edges, i.e., the total distance is minimum.
2.6 Linear regression
For relating the size with the variation of centrality, we employ the standard econometric technique of ordinary least squares. Let us assume that the model to explain datapoints, for , is given by the following
(5) 
where is the explanatory variable, is the dependent variable and denotes error terms. The ordinary least squares method minimizes the sum of the squared errors, to estimate the coefficients . Throughout our regression analyses, we have used scaled variables: [variablemean(variable)]/ standard deviation(variable). We carried out the estimation exercise using the MATLAB and R software packages.
3 Results
In this section, we describe the three main results. Firstly, the return correlation network closely mimics the actual production network and so the coreperiphery structure of the return correlation network is closely associated with the relative sizes of the sectors. Secondly, based on the findings for two periods of the empirical data (calm and volatile periods), our results are robust and hence universal, with respect to time. Finally, the core sectors which are typically very large in size, drives the market mode of the returns and hence, is riskier than the peripheral sectors as observed in minimum variance portfolio management.
3.1 Financial fluctuations and economic fundamentals
Given the return correlation , we computed the modified eigenvector centralities to find the core sectors of the countries, and to visualize the comovements and clusters of sectors based on return correlations, we applied two clustering algorithms, viz., MDS and MST. Fig. 1 (Upper) shows the MST. Fig. 1 (Upper Left Inset) shows that using the eigenvector centrality, we can identify that out of 10 sectors of the USA, 5 sectors constitute the core of the economy, viz., Finance (FN), Information Technology (IT), Industries (ID), Basic Materials (BM) and Consumer Discretionaries (CD) (see table 1 in Sec. 2.1 for names of the sectors). Fig. 1 (Lower Right Inset) shows the MDS. The MST generates a coreperiphery structure based on minimizing the distance between correlated sectors, and since it is a hierarchical clustering method, similar sectors can be found close to each other (or in one branch). Similarly, closer the sectors are placed on the MDS map, more correlated (similar) they are; farther they are placed on the map, less correlated they are.
There are two major observations: First, the MST shows that all core sectors form a chain or the “backbone” in the tree (see Fig. 1 (Upper)). Similarly, the MDS also reiterates the same information: the core sectors, as identified by the modified eigenvectors centrality, belong to one cluster in the MDS (see Fig. 1 (Lower Right Inset)); all sectors with negligible centrality are spaced in the periphery – far away from the core – in the MDS. Thus, our method of the modified centrality to extract the core sectors is reinforced by the clustering algorithms, indicating the robustness of our findings. Second, the MST built from the return correlation matrix, contains information about the actual production structure of the economy. For example, Energy (EG) is most closely related to Basic Materials (BM), which in turn is related to Industries (ID), and so on. On the other end of the MST, Consumer Staples (CS) is connected to Telecom (TC) sector, Utilities (UT) and Consumer Discretionary (CD). Again, this qualitative feature is quite robust, as observed in almost all the countries analyzed.
More importantly, we show that the coreperiphery structure based on the return correlation matrix, , has an intriguing relationship with the relative sizes of the sectors. In order to demonstrate and establish the relationship, we study the variations in the eigenvector centralities of the return correlation matrix, and exploit the variations in three major variables, viz., aggregate market capitalization, aggregate revenue and the aggregate employment. We have described in Sec. 2.1 how we constructed the sectorlevel data by aggregating the companylevel data. In Fig. 1 (Lower), we plot the linear regressions of scaled eigenvector centrality with the (scaled) market cap, revenue and employees for the USA. We have performed similar analyses for the other countries, and tabulated the results in the Supplementary material. Detailed analyses and tables suggest that generally, such a mapping exists for almost all countries.
Fig. 2 shows the coreperiphery structure for all countries. As can be seen, there are at least two sectors in the core for all countries, but the coreperiphery structure often changes with time (when compared for the periods 200809 and 201516). Thus, the relative importance of the sectors does change with time, and the sectoral dynamics and comovements may convey deeper insight about the aggregate macrolevel dynamics. In Fig. 3, we present similar MSTs (with the core/backbone colored in red) for 20 other countries, elucidating the coreperiphery structures.
Fig. 4 shows the results of regressing the sectoral eigenvector centralities on the sectorlevel aggregate market capitalization, revenue, and employees, for the years 200809 and 201516. As we see in Fig. 4 (Upper), for 201516, the coefficient for market capitalization 25 out of 27 countries are positive, and 11 out of those 27 countries have statistically significant relationships. The two countries which have very mildly negative relationships, are Greece (significant) and South Africa (insignificant). For 200809, the coefficient for 22 out of 26 countries are positive, and 3 out of those 26 countries have statistically significant relationships. Also, the countries Belgium, Switzerland, South Africa and Sri Lanka have negative relationships. In Fig. 4 (Middle), for 201516, the coefficient for revenue, 23 out of 26 countries are positive, and 9 out of those 26 countries have statistically significant relationships. The three countries which have negative (and statistically insignificant) relationships, are Greece, Qatar and United Kingdom. For 200809, the coefficient for 22 out of 26 countries are positive, and 9 out of those 24 countries have statistically significant relationships. Finally, in Fig. 4 (Lower), for 201516, the coefficient for employees, 23 out of 24 countries are positive, and 9 out of those 24 countries have statistically significant relationships. The only country which has negative (and statistically significant) relationship, is Greece. For 200809, the coefficient for 21 out of 24 countries are positive, and 5 out of those 24 countries have statistically significant relationships. For detailed statistical values of regressions performed on the sectorlevel aggregate data, please refer to the text in Supplementary material.
There is already an existing finding that centralities in inputoutput networks are closely related to the relative sizes of the corresponding nodes (see Ref. [9]). However, here we further show that the centralities based on nominal return fluctuations are related to relative size, i.e., the return network is also very closely related to the underlying size effects. An immediate corollary is that the core sectors of the return correlation network are also economically big, and hence, the market effect of the correlations are driven by the sectors, which have very high market capitalization (or other indicators like revenue and employment).
3.2 Robustness: volatile period versus calm period
We studied the dynamics of a total of 72 sectors across 27 countries, covering both developed and developing economies. Using methods of modified eigenvector centrality, MDS and MST, we can find the coreperiphery structure of all the economies. Fig. 2 showed the coreperiphery structure of all the countries, and indicated that most of the sectors do not change much in the coreperiphery structure during the periods of market turbulence, as well as relative calmness. There are of course, some sectors who were core in a volatile period, became the peripheral ones in the calm period, and vice versa. Fig. 5 shows the comparison among the modified eigenvector centralities for the years 200809 and 201516, for the four countries: United Kingdom, India, Japan, and United States of America, as examples.
The relative importance of each sector can be compared for the volatile and calm period. Certainly the sectoral dynamics are interesting to note in the different countries, and may help in taking important policy decisions in economic growth and development.
3.3 Constructing the minimum risk portfolio
In this part, we study how the sectoral centralities influence the aggregate risk of a portfolio. For the purpose of simple exposition, we compute the benchmark model of Markowitz portfolio selection with the sectoral return data. Assuming rational investors with riskaversion, the investors will minimize
(6) 
with respect to the weight vector , where is the covariance matrix of the sectoral returns, is the expected return vector and is a parameter which denotes the risk appetite of the investor. We set a shortselling constraint ( 0) and equals to zero for finding the minimal risk portfolio which will entail a convex combination of sectoral returns (the other extreme would lead to a corner solution).
Our main observation in this part is that the optimal weight vector, , is negatively related to the eigenvector centralities, i.e., if a sector is very “central” in the return correlation network , then it is less likely to appear in the optimal portfolio with the minimum risk (and no short selling). We demonstrate this in a naive way: we construct threshold values and , as a fixed percentage (say of the coefficient of variation (standard deviation/mean) for both the eigenvector centralities, as well as the minimum risk portfolio weights, respectively. These threshold values and would determine, respectively, whether the sector is central or not (i.e., or ), or whether the corresponding sector would appear in your optimal portfolio or not (i.e., or ). So, for the vector of sectors, we would have two strings of ’s or ’s corresponding to the centrality vector (EVC) and the optimal weight vector (PWT), respectively. The Hamming distance between any two bitstrings of equal length, is the number of positions at which the corresponding bits are different. So, the Hamming distance between the two strings EVC and PWT would tell how significant the observation is for a particular country; higher the value of , better the conformity. The sector which is central (i.e., ) would not appear in your portfolio (i.e., ), and so for any country the ideal finding would be that is unity. The choice of the threshold(s), and , equaled by the percentage(s) () of the coefficient of variation(s) in the vectors EVC and PWT, would be important for determining the Hamming distance between the strings for any country (see Fig. 6 (Upper)) for the USA. We can optimize the value of against the percentage , for all the countries, as shown in Fig. 6 (Lower). We found that , i.e. 2% was an optimal threshold value for most countries, which we then used to distinguish between the core and periphery sectors. Combined with the finding that core sectors in the return correlation network are bigger in size, the above finding implies that peripheral sectors contribute to lower risk of a diversified portfolio.
4 Summary and conclusion
In this paper, we have analyzed financial and economic data for 27 countries at the sector level. We show that the variation in the centrality in the return correlation matrix across sectoral indices, can be explained by the size dispersion across the sectors. This finding indicates that financial fluctuations are mapped to the macroeconomic fundamentals. From the perspective of portfolio optimization, we show that the very big sectors that are also highly central in the return network, rarely appear in a riskminimizing portfolio. Essentially, such sectors constitute the main drivers of the marketwide fluctuations. In summary, our study sheds light on: (a) the mapping between the joint evolution of the financial variables and the underlying macroeconomic fundamentals, and (b) extracting information about the individual influences on aggregate risk from sectorlevel, disaggregated timeseries data.
Methodologically, we provide a way to extract the coreperiphery structure of the correlation networks in a binary fashion. As a result, the generic rule of thumb we come up with is that size is an important causal factor even behind financial fluctuations. We attribute significant importance to this finding as it provides a way to exactly pin down the sectors, which are main drivers of financial fluctuations through the size effect. The way return series are constructed, the size differential of the prices across the sectoral indices, should disappear due to the normalization. The fact that the comovements are still tied to the fundamentals is therefore intriguing. As our results suggest, the finding is considerably robust across countries. An illuminating exception is Greece showing an exact opposite relationship, which has been known to possess weak economic fundamentals along with severe crises in the financial markets in the recent times. In both periods, economically large (either in terms of market capitalization or revenue or employment) sectors in Greece are at the periphery of the return correlation networks, which constitute an inverted relationship between the economy and the financial networks.
We have also shown that the relative importance of the sectors may change significantly over time although some sectors like finance and industry are at the core of a significant fraction of countries. In general, our results indicate that the core may not be very stable. Possible reasons could be sectoral competition in terms of productivity and innovation and the resultant evolution [17]. The emergence of the coreperiphery structure changes the complexity of the financial markets and has implications of the pricing of risk in the economy [18]. Our work indicates the potentials of using a binary characterization to reduce the computational burden by introducing proper identification of the countryspecific core sectors, as opposed to considering the full network.
To conclude, we note that the recent applications of network theory in the macroeconomics literature has focused mostly on studying the dynamics of real economic quantities [19], whereas the relevant finance literature has focused on the dynamics of nominal quantities [20]. The present work may provide a linkage between the two. In other words, we make the point that the oftquoted quips ‘toobigtofail’ and ‘toointerconnectedtofail’ may not be as different as is currently thought of [21].
Acknowledgment
ASC acknowledges the support by the institute grant (R&P), IIM Ahmedabad. BG acknowledges FPM fellowship provided by IIM Ahmedabad. AC and KS acknowledge the support by grant number BT/BI/03/004/2003(C) of Govt. of India, Ministry of Science and Technology, Department of Biotechnology, Bioinformatics division, and University of Potential ExcellenceII grant (Project ID47) of the Jawaharlal Nehru University, New Delhi. KS acknowledges the University Grants Commission (Ministry of Human Research Development, Govt. of India) for her senior research fellowship.
Supplementary material
All detailed regression tables are provided below.
Countries  Tstat  Pvalue  Tstat  Pvalue  Rsquare  
Australia 

0.0000  0.9999 

0.5773  0.5817  0.0454  
Belgium 

0.0000  0.9999 

0.6371  0.5475  0.0633  
Canada 

0.0000  0.9999 

0.2404  0.8148  0.0057  
Denmark 

0.0000  0.9999 

2.0767  0.0714*  0.3502  
Finland 

0.0000  0.9999 

3.1353  0.0139**  0.5513  
France 

0.0000  0.9999 

2.7751  0.0241**  0.4904  
Germany 

0.0000  0.9999 

2.4299  0.0412**  0.4246  
Greece 

0.0000  0.9999 

5.1368  0.0008***  0.7673  
Hong Kong 

0.0000  0.9999 

1.3123  0.2257  0.1771  
India 

0.0000  0.9999 

0.3474  0.7373  0.0149  
Indonesia 

0.0000  0.9999 

2.9524  0.0183**  0.5214  
Japan 

0.0000  0.9999 

0.7579  0.4702  0.0670  
Malaysia 

0.0000  0.9999 

2.0421  0.0754*  0.3426  
Netherlands 

0.0000  0.9999 

1.0941  0.3100  0.1460  
Norway 

0.0000  0.9999 

1.4316  0.1901  0.2039  
Philippines 

0.0000  0.9999 

5.4929  0.0118**  0.9095  
Portugal 

0.0000  0.9999 

1.7061  0.1388  0.3266 
Qatar 

0.0000  0.9999 

1.3959  0.2352  0.3275  
Saudi Arabia 

0.0000  0.9999 

1.0133  0.3309  0.0788  
South Africa 

0.0000  0.9999 

0.5086  0.6291  0.0413  
Spain 

0.0000  0.9999 

0.8057  0.1577  0.0234  
Sri Lanka 

0.0000  0.9999 

1.9363  0.0848*  0.2940  
Sweden 

0.0000  0.9999 

1.9818  0.0879*  0.3594  
Switzerland 

0.0000  0.9999 

1.5302  0.1603  0.2064  
Thailand 

0.0000  0.9999 

1.6743  0.1450  0.3184  
UK 

0.0000  0.9999 

1.0830  0.3103  0.1278  
USA 

0.0000  0.9999 

2.0228  0.0777*  0.3384 
Countries  Tstat  Pvalue  Tstat  Pvalue  Rsquare  
Australia 

0.0000  0.9999 

1.4106  0.2012  0.2213  
Belgium 

0.0000  0.9999 

3.3644  0.0151**  0.6535  
Canada 

0.0000  0.9999 

0.8830  0.3979  0.0723  
Denmark 

0.0000  0.9999 

1.7834  0.1123  0.2844  
Finland 

0.0000  0.9999 

1.4422  0.1872  0.2063  
France 

0.0000  0.9999 

2.8557  0.0212**  0.5048  
Germany 

0.0000  0.9999 

0.8778  0.4056  0.0878  
Greece 

0.0000  0.9999 

1.2453  0.2482  0.1623  
Hong Kong 

0.0000  0.9999 

1.6948  0.1285  0.2642  
India 

0.0000  0.9999 

1.0117  0.3413  0.1134  
Indonesia 

0.0000  0.9999 

2.9286  0.0190**  0.5173  
Japan 

0.0000  0.9999 

1.0678  0.3167  0.1247  
Malaysia 

0.0000  0.9999 

1.9824  0.0827*  0.3294  
Netherlands 

0.0000  0.9999 

0.7976  0.4512  0.0833  
Norway 

0.0000  0.9999 

3.2031  0.0125**  0.5618  
Philippines 

0.0000  0.9999 

0.9035  0.4328  0.2139  
Portugal 

0.0000  0.9999 

2.5320  0.0445**  0.5165 
Qatar 

0.0000  0.9999 

0.2219  0.8352  0.0121  
Saudi Arabia 


South Africa 

0.0000  0.9999 

1.7121  0.1377  0.3282  
Spain 

0.0000  0.9999 

0.2767  0.7999  0.0248  
Sri Lanka 

0.0000  0.9999 

2.5267  0.0324**  0.4150  
Sweden 

0.0000  0.9999 

1.8006  0.1147  0.3165  
Switzerland 

0.0000  0.9999 

1.9952  0.0771*  0.3066  
Thailand 

0.0000  0.9999 

0.9538  0.3770  0.1316  
UK 

0.0000  0.9999 

0.1952  0.8500  0.0047  
USA 

0.0000  0.9999 

2.1566  0.0631*  0.3676 
Countries  Tstat  Pvalue  Tstat  Pvalue  Rsquare  
Australia 

0.0000  0.9999 

1.6213  0.1489  0.2730  
Belgium 

0.0000  0.9999 

2.5766  0.0419**  0.5252  
Canada 

0.0000  0.9999 

0.8698  0.4048  0.0703  
Denmark 

0.0000  0.9999 

0.9942  0.3492  0.1099  
Finland 

0.0000  0.9999 

1.9474  0.0873*  0.3216  
France 

0.0000  0.9999 

5.5374  0.0005***  0.7930  
Germany 

0.0000  0.9999 

1.7845  0.1121  0.2847  
Greece 

0.0000  0.9999 

5.7451  0.0004***  0.8049  
Hong Kong 

0.0000  0.9999 

2.5943  0.0318**  0.4569  
India 

0.0000  0.9999 

0.0534  0.9588  0.0003  
Indonesia 

0.0000  0.9999 

1.8283  0.1049  0.2947  
Japan 

0.0000  0.9999 

1.5982  0.1486  0.2420  
Malaysia 

0.0000  0.9999 

1.4717  0.1792  0.2130  
Netherlands 

0.0000  0.9999 

3.1794  0.0155  0.5908  
Norway 

0.0000  0.9999 

2.0054  0.0798*  0.3345  
Philippines 

0.0000  0.9999 

0.7996  0.4823  0.1756  
Portugal 

0.0000  0.9999 

1.5820  0.1647  0.2943 
Qatar 


Saudi Arabia 


South Africa 

0.0000  0.9999 

0.4008  0.7024  0.0260  
Spain 

0.0000  0.9999 

0.6550  0.5591  0.1251  
Sri Lanka 

0.0000  0.9999 

3.9435  0.0033***  0.6334  
Sweden 

0.0000  0.9999 

1.3986  0.2046  0.2184  
Switzerland 

0.0000  0.9999 

1.8154  0.1028  0.2680  
Thailand 


UK 

0.0000  0.9999 

2.4165  0.0420**  0.4219  
USA 

0.0000  0.9999 

2.4628  0.0391**  0.4312 
Countries  Tstat  Pvalue  Tstat  Pvalue  Rsquare  
Australia 

0.0000  0.9999 

0.4759  0.6485  0.0313  
Belgium 

0.0000  0.9999 

0.2943  0.7783  0.0142  
Canada 

0.0000  0.9999 

0.0001  0.9998  0.0000  
Denmark 

0.0000  0.9999 

1.2727  0.2388  0.1683  
Finland 


France 

0.0000  0.9999 

1.2329  0.2526  0.1596  
Germany 

0.0000  0.9999 

2.0484  0.0746*  0.3440  
Greece 

0.0000  0.9999 

0.3587  0.7290  0.0158  
Hong Kong 

0.0000  0.9999 

0.9684  0.3611  0.1049  
India 

0.0000  0.9999 

1.9605  0.0855*  0.3245  
Indonesia 

0.0000  0.9999 

1.5861  0.1513  0.2392  
Japan 

0.0000  0.9999 

0.2600  0.8013  0.0083  
Malaysia 

0.0000  0.9999 

2.0188  0.0782*  0.3375  
Netherlands 

0.0000  0.9999 

1.3686  0.2134  0.2110  
Norway 

0.0000  0.9999 

1.3739  0.2067  0.1909  
Philippines 

0.0000  0.9999 

1.2592  0.2970  0.3457  
Portugal 

0.0000  0.9999 

1.7346  0.1334  0.3340 
Qatar 

0.0000  0.9999 

1.1987  0.2967  0.2642  
Saudi Arabia 

0.0000  0.9999 

0.7734  0.4542  0.0474  
South Africa 

0.0000  0.9999 

0.2293  0.8262  0.0086  
Spain 

0.0000  0.9999 

0.5928  0.5950  0.1048  
Sri Lanka 

0.0000  0.9999 

0.7120  0.4945  0.0533  
Sweden 

0.0000  0.9999 

0.7297  0.4892  0.0706  
Switzerland 

0.0000  0.9999 

0.1599  0.8764  0.0028  
Thailand 

0.0000  0.9999 

0.6845  0.5191  0.0724  
UK 

0.0000  0.9999 

0.1959  0.8495  0.0047  
USA 

0.0000  0.9999 

0.4315  0.6774  0.0227 
Countries  Tstat  Pvalue  Tstat  Pvalue  Rsquare  
Australia 

0.0000  0.9999 

0.9592  0.3694  0.1161  
Belgium 

0.0000  0.9999 

0.6891  0.5164  0.0733  
Canada 

0.0000  0.9999 

1.0658  0.3115  0.1020  
Denmark 

0.0000  0.9999 

1.6254  0.1427  0.2482  
Finland 


France 

0.0000  0.9999 

2.6273  0.0303**  0.4631  
Germany 

0.0000  0.9999 

1.8814  0.0966*  0.3067  
Greece 

0.0000  0.9999 

1.2401  0.2500  0.1612  
Hong Kong 

0.0000  0.9999 

1.5142  0.1684  0.2227  
India 

0.0000  0.9999 

1.3601  0.2108  0.1878  
Indonesia 

0.0000  0.9999 

1.9798  0.0830*  0.3288  
Japan 

0.0000  0.9999 

0.8018  0.4458  0.0743  
Malaysia 

0.0000  0.9999 

1.9555  0.0862*  0.3234  
Netherlands 

0.0000  0.9999 

0.8122  0.4433  0.0861  
Norway 

0.0000  0.9999 

2.5688  0.0331**  0.4520  
Philippines 

0.0000  0.9999 

0.6959  0.5365  0.1389  
Portugal 

0.0000  0.9999 

1.9705  0.0962*  0.3928 
Qatar 

0.0000  0.9999 

1.1979  0.2970  0.2640  
Saudi Arabia 

0.0000  0.9999 

0.6064  0.5555  0.0297  
South Africa 

0.0000  0.9999 

1.6100  0.1585  0.3017  
Spain 

0.0000  0.9999 

1.7139  0.1850  0.4947  
Sri Lanka 

0.0000  0.9999 

0.2579  0.8022  0.0073  
Sweden 

0.0000  0.9999 

1.5566  0.1634  0.2571  
Switzerland 

0.0000  0.9999 

0.3765  0.7152  0.0155  
Thailand 

0.0000  0.9999 

0.4761  0.6507  0.0364  
UK 

0.0000  0.9999 

0.5427  0.6020  0.0355  
USA 

0.0000  0.9999 

0.4091  0.6931  0.0205 
Countries  Tstat  Pvalue  Tstat  Pvalue  Rsquare  
Australia 

0.0000  0.9999 

0.9095  0.3932  0.1056  
Belgium 

0.0000  0.9999 

0.2259  0.8287  0.0084  
Canada 

0.0000  0.9999 

0.7154  0.4907  0.0486  
Denmark 

0.0000  0.9999 

1.0226  0.3364  0.1156  
Finland 


France 

0.0000  0.9999 

2.3971  0.0433**  0.4180  
Germany 

0.0000  0.9999 

1.6066  0.1468  0.2439  
Greece 

0.0000  0.9999 

0.0502  0.9611  0.0003  
Hong Kong 

0.0000  0.9999 

1.7222  0.1233  0.2704  
India 

0.0000  0.9999 

1.1991  0.2647  0.1523  
Indonesia 

0.0000  0.9999 

1.9759  0.0835*  0.3279  
Japan 

0.0000  0.9999 

1.5575  0.1579  0.2326  
Malaysia 

0.0000  0.9999 

1.7870  0.1117  0.2853  
Netherlands 

0.0000  0.9999 

1.3679  0.2136  0.2109  
Norway 

0.0000  0.9999 

2.6034  0.0314**  0.4586  
Philippines 

0.0000  0.9999 

0.6777  0.5465  0.1327  
Portugal 

0.0000  0.9999 

1.4425  0.1992  0.2575 
Qatar 


Saudi Arabia 


South Africa 

0.0000  0.9999 

0.7753  0.4675  0.0910  
Spain 

0.0000  0.9999 

1.3707  0.2640  0.3851  
Sri Lanka 

0.0000  0.9999 

1.0594  0.3170  0.1109  
Sweden 

0.0000  0.9999 

1.2679  0.2453  0.1867  
Switzerland 

0.0000  0.9999 

0.7481  0.4734  0.0585  
Thailand 

0.0000  0.9999 

0.8259  0.4404  0.1020  
UK 

0.0000  0.9999 

2.6571  0.0289**  0.4688  
USA 

0.0000  0.9999 

1.9171  0.0915*  0.3148 
References
 R. Shiller, “Do stock prices move too much to be justified by subsequent changes in dividends?,” American Economic Review, pp. 421–436, 1981.
 X. Gabaix, “Variable rare disasters: an exactly solved framework for ten puzzles in macrofinance,” The Quarterly Journal of Economics, vol. 127, pp. 645–700, 2012.
 D. Sornette, Why stock markets crash: critical events in complex financial systems. Princeton Univeristy Press, 2004.
 M. Newman, Networks: an introduction. Oxford university press, 2010.
 R. Axtell, “Zipf distribution of U.S. firm sizes,” Science, vol. 293, pp. 1818–1820, 2001.
 M. H. R. Stanley, L. A. N. Amaral, S. V. Buldyrev, S. Havlin, H. Leschhorn, P. Maass, M. A. Salinger, and H. E. Stanley, “Scaling behaviour in the growth of companies,” Nature, vol. 379, pp. 804–806, 1996.
 X. Gabaix, “The granular origins of aggregate fluctuations,” Econometrica, vol. 79, pp. 733–772, 2011.
 Y. Lee, L. A. N. Amaral, M. Meyer, D. Canning, and H. E. Stanley, “Universal features in the growth dynamics of complex organizations,” Physical Review Letters, vol. 81, pp. 3275–3278, 1998.
 D. Acemoglu, V. M. Carvalho, A. Ozdaglar, and A. TahbazSalehi, “The network origin of aggregate economic fluctuations,” Econometrica, vol. 80, no. 5, pp. 1977–2016, 2012.
 E. Atalaya, A. Hortaçsu, J. Roberts, and C. Syverson, “Network structure of production,” Proceedings of the National Academy of Sciences, vol. 108, no. 13, pp. 5199–5202, 2011.
 M. Jackson, Social and economic networks. Princeton Univeristy Press, 2010.
 Thompson Reuters Eikon Database 2016. Accessed on 9th November, 2016.
 R. Rammal, G. Toulouse, and M. A. Virasoro, “Ultrametricity for physicists,” Review of Modern Physics, vol. 58, p. 765, 1986.
 R. N. Mantegna, “Hierarchical structure in financial markets,” European Physical Journal B, vol. 11, p. 193, 1999.
 J.P. Onnela, A. Chakraborti, K. Kaski, J. Kertesz, and A. Kanto, “Dynamics of market correlations: taxonomy and portfolio analysis,” Physical Review E, vol. 68, p. 056110, 2003.
 I. Borg and P. Groenen, “Modern multidimensional scaling: theory and applications,” SpringerVerlag, New York, 2005.
 D. Acemoglu, U. Akcigit, and W. R. Kerr, “Innovation network,” Proceedings of the National Academy of Sciences, vol. 113, no. 41, pp. 11483–11488, 2016.
 S. Battiston, G. Caldarelli, R. M. May, T. Roukny, and J. E. Stiglitz, “The price of complexity in financial networks,” Proceedings of the National Academy of Sciences, vol. 113, no. 36, pp. 10031–10036, 2016.
 “Crises and complexity,” Special issue in Journal of Economic Dynamics and Control, vol. 50, pp. 1–202, 2015.
 R. N. Mantegna and H. E. Stanley, Introduction to Econophysics: correlations and complexity in finance. Cambridge University Press, 2007.
 D. Acemoglu, A. Ozdaglar, and A. TahbazSalehi, “Systemic risk and stability in financial networks,” American Economic Review, vol. 105 (2), pp. 564–608, 2015.