Allianz Research Digital-enabling Countries Proved More

Preparing to load PDF file. please wait...

0 of 0
100%
Allianz Research Digital-enabling Countries Proved More

Transcript Of Allianz Research Digital-enabling Countries Proved More

ALLIANZ RESEARCH
DIGITAL-ENABLING COUNTRIES PROVED MORE RESILIENT TO THE COVID-19 ECONOMIC SHOCK

17 February 2021
ALEXIS GARATTI Global Head of Economic Research [email protected] GEORGES DIB Economist [email protected] LOUIS ADJIMAN Research Assistant [email protected]

The US, Germany and Denmark once again make the top three of our 2020 Enabling Digitalization Index (based on data from end-2019). The EDI measures the ability – and agility – of countries to help digital companies thrive and traditional businesses harness the digital dividend. It scores 115 countries based on five components: regulation, knowledge, connectivity, infrastructure and size. For 2020, the US leads by far due to its best-in-class knowledge ecosystem, competitive market size and favorable regulation. In fact, its connectivity score has increased by +1.8 points after a +5.1 point increase in 2018 (see Appendix 1). Meanwhile, Germany boasts the best knowledge ecosystem and infrastructure for trade. It saw a moderate improvement in both the regulation and market size scores, but its connectivity quality has dropped relative to the rest of the world despite the continuing upwards trend in the number of secure servers. This is due to fewer mobile lines per 100 inhabitants and a slightly declining share of internet users. Denmark started 2020 as the best performer in terms of connectivity quality. Indeed, after tripling its number of secure servers in 2018, it has more than doubled it again to reach a higher number than China and Canada, and close to that of France (with a population of only 6 million).
China’s rise seems unstoppable. In the three years preceding the outbreak of Covid-19, China moved from rank 17 to rank 4. China has seen rising scores across the board: the country’s regulation score improved by +7.4 points after increasing by +15 points in 2018. The connectivity score also increased by +1.3 points. Lastly, the knowledge score rose by +12 points due to an increase in China’s innovation capability over 2019. Yet, the skills score did not follow the same pattern, highlighting that China still has leeway to boost the skills (especially digital skills) of its population. This would allow Chinese companies to appropriately tap its innovation potential.
Data also show that others in Asia made progress in the years preceding the Covid-19 crisis: Hong Kong, now at rank 7, previously 11. South Korea, at rank 12 up from rank 16. Six out of the fifteen top digital enablers were in the Asia-Pacific region at the end of 2019. France had also advanced by two spots to rank 15, and Spain had gained 4 spots to rank 20. Other remarkable progressions include Vietnam from 67 to 57 and Saudi Arabia from 53 to 41, confirming a clear willingness to transition towards a new model of growth.

1

Our estimates1 show that an additional point in a country’s 2020 EDI score translated to +0.25pp GDP growth in Q3 2020 y/y (i.e. compared to Q3 2019), suggesting that digitization plays the role of shock absorber. The economic interpretation is that countries whose environment was more conducive to the digitalization of companies (good connectivity, market size, regulation, logistics and knowledge) were likely able to respond to the crisis by ramping up multidimensional digitalization. Those countries likely enabled digitalization in administrative bureaucratic processes (state schemes to help companies and citizens in rapidly receiving financial help or sanitary assistance (testing, tracing, isolating, distributing vaccines), on the demand side (consumption with the help of web platforms) and on the supply side in terms of companies’ ways of working (remote working, data storing and sharing etc.).

Our estimates also point to a statistically significant relationship between a country’s economic performance in 2020 and the share of services in its value added, as well as the widening of its public deficit. This confirms that service-oriented economies with prominent arts, recreation, restaurants, hotels, and other tourism-related sectors suffered relatively more, all other things equal. As for the widening of the public deficit, it appears that higher spending was associated with a higher hit to the economy. This is probably due to the severity of lockdown measures; countries that were the most aggressive in closing their economies to control the pandemic would have had to resort to offsetting actions on the fiscal side to absorb the shock. This is confirmed by the fact that that the stringency of government measures to fight the pandemic was not significantly correlated to a country’s economic performance in 2020 across all 78 countries (size of the deficit capturing the information of this variable), nor were bureaucracy quality, institutional effectiveness, power distance and the democracy index2.

TableTa1bl:eM1 :oLidneealrrreegsruesltssion of GDP growth (Q3 2020, Yoy) on centered mentioned variables

Model 1

EDI

0.25***

(0.05)

Stringency

-0.12 (p-value=0.053)

(0.06)

Share of services

-0.36***

(0.073)

Public Deficit Widening

-0.56**

(0.17)

*Dpem
Bureaucracy Quality

Model 2 0.25***
(0.048)
-0.10
(0.06)
-0,39***
(0.08)
-0.58**
(0.17)
0.29
(0.35)

Model 3 0.23*
(0.07)
-0.07
(0.08)
-0.24*
(0.04)
-0.48*
(0.20)
-0.12
(0.37)
-0.34
(0.92)

Grouping countries across our reg* pr
1 In order to investigate the link between the Enabling Digitalization Index and resilience to the economic shock, we regressed Q3 GDP growth (y/y, i.e. compared to Q3 2019) against several variables across 78 countries. The explanatory variables were the following: EDI (a high EDI means a high potential for digitalization), Oxford’s Stringency Index (the higher, the more stringent), the change in the primary fiscal balance between 2020 and 2019 (an increase corresponds to a widening of the fiscal deficit) in % of GDP and the share of services in total added value in 2019 (%). All the variables were taken as a deviation compared with the average of the sample. An estimate between Q4 2019 and Q4 2020 did not drastically change our results. 2 See methodology in the Appendix for more details.
2

granularity and precision by splitting it into two groups, according to the same criteria. It became apparent that even though the two groups presented comparable GDP reactions to the crisis, one was much more digitalized, stringent and with a bigger share in services than the other. This first subgroup of countries was much more exposed to the crisis (since they adopted heavier stay-at-home measures and relied much more on services, the most Covid-sensitive part of the economy), and yet they managed to limit economic losses in the same way as the second subgroup, thanks to their high potential for digitalization. In this group we find the US, Denmark, Germany, China, the UK, Singapore, Switzerland, Sweden, Austria, France, Finland, Australia, France, Belgium, Spain, and Luxembourg.
The second cluster is the “median” one, comprising countries with median values of each variable. This list is diverse, including some Western European and Latin American countries, which fared worse in terms of GDP performance (-8%).
The third cluster comprises those most hit by the Covid-19 crisis, and those with the lowest EDI and low fiscal spending in 2020 compared to 2019: mainly Latin American and Middle Eastern countries. Their mean economic performance is -9.4% y/y in Q3. These countries also implemented the most stringent restriction measures, and were forced to use fiscal policy to absorb the shock.
The last cluster comprises countries where a disaster has been avoided: they have low EDI and low fiscal spending but faced a benign economic shock (-2% y/y in Q3) relative to the rest of the sample. Most of these countries are located in Africa, where the pandemic did not spread as widely as in the rest of the world: activity was not halted by government measures, and the state did not resort to overly large fiscal spending programs to save the economy.
Lastly, we test the robustness of our EDI indicator and regression by replacing it with other digitalization-related indices. We obtain similar results, with digitalization being significantly positively correlated to economic performance.
Policymakers have another incentive to boost digitalization: Companies reporting higher digital adoption proved more resilient and better prepared for future challenges amid the Covid-19 crisis. Our Global Supply Chain Survey3 showed that digitalization correlates with resilience. In fact, digitalization means agility and proactivity: highly digitized companies took more swift action to mitigate the supply-chain disruptions in 2020. In contrast, low digitization was synonymous with indecision: 35% of less digitized companies neither agreed nor disagreed when we asked them if the pandemic would push them to find new suppliers, double the share of those with the same response among highly digitized companies.
Highly digitized companies are also more forward looking, and this contributes to resilience. Digitalization means better knowledge and preparedness for the future: 80% of mostly highly digitized companies
3 We surveyed a sample of high-level executives in 1,181 companies in these countries across six sectors (IT, tech and telecoms, machinery and equipment, chemicals, energy and utilities, automotive and agrifood) about their experiences with disruption and their plans to make their supply chains more resilient. The survey was conducted online from mid-October to early November.
3

know their Tier-2 suppliers vs. 61% of the less digitized. As these companies are faced with more diverse and complex risks, they also have a better information management system and better risk analysis. 25% of less digitized companies stated they would prefer a local supply chain but their market or company cannot afford a cost increase vs. 9% of highly digitized companies. Highly digitized companies are perhaps more non-cost competitive, while the other are cost-competitive. Our survey also shows that digitalization can mean exposure to more complex and multidimensional risks. Highly digitized companies see protectionism as a higher risk (11%) for supply chains than companies with a low level of digitalization (4% of respondents only) and those with a medium level (6%). Political risk on the supply chain matters much more for the most digitized companies than for those who have a lower level of digitalization (31% vs. 13%). ESG risk to production sites was also more important for highly digitized companies (26% vs. 12% for those with low level of digitalization).
4

APPENDIX 1 : Top 40 Countries in EDI 2020

Global ranking Country

Connectivity quality

1

US

76

2

Denmark

100

3

Germany

76

4

China

34

5

UK

70

6

Singapore

80

7

Hong Kong

91

8

Japan

66

9

Netherlands

80

10

Switzerland

76

11

Sweden

59

12

South Korea

68

13

Austria

64

14

New Zealand

64

15

France

68

16

Finland

57

17

Canada

62

18

Australia

60

19

Belgium

60

20

Spain

64

21

UAE

66

22

Norway

57

23

Ireland

70

24

Iceland

75

25

Luxembourg

72

26

Italy

54

27

Estonia

70

28

Israel

60

29

Czech Republic

55

30

Portugal

60

31

Poland

54

32

Slovenia

62

33

Malaysia

53

34

Hungary

56

35

Lithuania

56

36

Qatar

57

37

Thailand

42

38

Russia

55

39

Cyprus

61

40

Chile

50

Logistic performance
86 90 100 72 90 91 87 92 92 86 93 72 92 85 83 89 78 79 93 83 89 77 68 55 73 79 58 58 76 74 69 58 54 64 45 66 63 33 51 59

Business environment
95 97 87 84 94 99 97 84 81 82 92 95 86 100 82 88 87 90 79 84 90 93 87 86 70 75 89 82 81 82 82 82 91 76 91 68 88 85 76 75

Knowledge Ecosystem
97 95 100 75 94 90 82 89 94 98 95 90 90 83 87 94 91 88 88 79 72 89 83 87 86 79 77 90 76 72 71 78 75 68 72 70 62 71 69 66

Market Size
83 1 18 100 13 1 2 23 4 3 2 9 2 1 13 1 8 6 2 8 2 2 1 0 0 10 0 2 1 1 4 0 3 1 0 1 5 14 0 2

EDI (end-2019 data)
88 77 76 73 72 72 72 71 70 69 68 67 67 67 67 66 65 65 64 64 64 63 62 60 60 59 59 58 58 58 56 56 55 53 53 52 52 52 52 50

Ranking Change vs. end-2018
0 1 -1 5 0 0 4 0 -5 -3 -1 4 -1 0 2 -3 -2 0 0 4 2 -2 -2 -2 0 2 -1 1 -2 0 1 -1 0 0 0 0 3 -1 -1 -1

APPENDIX 2: METHODOLOGY
Table 1 : Linear regression of GDP growth (Q3 2020, Yoy) on centered mentioned variables

EDI Stringency Share of services Public Deficit Widening Democracy Index Bureaucracy Quality

Model 1 0.25***
(0.05)
-0.12 (p-value=0.053)
(0.06)
-0.36***
(0.073)
-0.56**
(0.17)

Model 2 0.25***
(0.048)
-0.10
(0.06)
-0,39***
(0.08)
-0.58**
(0.17)
0.29
(0.35)

Model 3 0.23*
(0.07)
-0.07
(0.08)
-0.24*
(0.04)
-0.48*
(0.20)
-0.12
(0.37)
-0.34
(0.92)

* p<0.05 ; ** p< 0.01 ; *** p< 0.001

Table 1 shows the coefficients obtained when regressing our GDP variable on several variables. In each model, we add another variable to account
5

for more and more effects. The coefficient associated to EDI is always at least 95% significant, which is all the more reason to trust the positive effect of being a digital-enabling country. This is also the case for the share of services and deficit variables, for which the coefficients are always negative, which makes sense for the former: the more an economy relied on services, the harder it was hit by the pandemic crisis.
However, the sign of the deficit coefficient is more ambiguous. A negative sign indicates that the more a country widened its deficit to fight the pandemic, the bigger the GDP loss between Q3 2020 and Q3 2019. Yet we know that increasing the deficit allows a country to absorb part of the damage done to the economy and hereby reduce GDP loss. The answer probably lies in the fact that countries that faced the crisis the hardest are also the ones that had to provide the biggest budget effort. In other terms, there is reverse causality here, which means that GDP loss has caused a rise in the fiscal deficit as much as the deficit prevented GDP from going any lower. In such cases, the concerned regression estimate is often heavily biased and inconsistent.
A surprising result is that our estimates did not allow for an unquestionable assessment of stringent measures, as Model 1 (the simplest) was the only model in which the coefficient associated with the stringency variable was 94% significant. In the two other models, significance was uncertain, which may seem odd since one can have the feeling that stringent political measures have driven GDP loss. The information contained in this variable can actually be captured by the deficit variable, as stricter confinements were often accompanied with higher deficits (accompanying measures).
Coefficients associated with the Democracy Index and Bureaucracy Quality, which aimed at capturing the effectiveness with which a government could implement sanitary decisions such as vaccination campaigns or stay-at-home measures, never proved significant. Rather, their inclusion seemed to lower other variables’ significance. Several other indicators of state centrality were used, such as Institutional Effectiveness and Policy Implementation (OECD), but the results were somewhat poorer in terms of coefficient significance.
Once we establish positive effect of the EDI on GDP resilience, we try to identify different groups of countries with respect to the aforementioned variables. These groups will be visualized, thanks to a previously executed PCA, which allows us to identify and interpret the main axes on which countries will be projected.
6

This PCA shows how all

five variables contribute to

the formation of each

component. In total, the

two axes explain

42.36+25.44=67.8% of the

total dataset variance.

Since

each

PCA

component is a linear

combination of the

different variables, the

coordinates of each variable gives its coefficient in the linear relation that

forms each component. For example, we have:

𝐶𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡1 = 0,46. 𝑆𝑡𝑟𝑖𝑛𝑔𝑒𝑛𝑐𝑦 + 0. 𝐺𝐷𝑃 − 0,88. 𝐸𝐷𝐼 − 0,82. 𝑆𝑒𝑟𝑣𝑖𝑐𝑒𝑠 𝑆ℎ𝑎𝑟𝑒 − 0,7. 𝐵𝑎𝑙𝑎𝑛𝑐𝑒 𝐶ℎ𝑎𝑛𝑔𝑒

We can therefore see that Component 1 is mainly composed of EDI, share of services and balance change, while GDP is the main contribution to Component 2. These observations will facilitate our clustering interpretations.

A k-means clustering on all variables allows us to automatically assign each country to a certain group to minimize the variance within each group, that is, to produce groups that are coherent and as compact as they can be. Our tests led us to choose four groups, since adding another one would not have improved our results enough to compensate for the loss in visibility. With every country belonging to a certain group, we can now project our data on the previously constructed PCA components to try and visualize them (the country repartition details are in the Appendix).

On the projection below, each dot represents a country and its color the cluster it belongs to. The “X” letters stand for the centers of each cluster. Notice those are the same axes as the PCA’s, which means we could also draw the arrows representing each variable from the previous factor map. From there, we can characterize each group.

7

Then, a second k-means solely on cluster 0 was executed to form the list of countries available below.

Countries in the different clusters :

Cluster 0

Sub-group 1 Sub-group 2

United States Netherlands

Denmark

UAE

Germany

Norway

China

Ireland

United Kingdom Iceland

Singapore

Italy

Switzerland Estonia

Sweden

Czech Republic

Austria

Portugal

France

Poland

Finland

Slovenia

Australia

Hungary

Belgium

Lithuania

Spain

Russia

Luxembourg Cyprus

Brazil

Cluser 2 Thailand Greece Latvia Croatia Malta Romania Bulgaria Mexico Costa Rica Uruguay Botswana Tunisia Namibia

Cluster 1 Malaysia Qatar Saudi Arabia India Turkey Bahrain Indonesia Ukraine Egypt Kenya Paraguay Ghana Nigeria Pakistan Senegal Cameroon Mali Mozambique Ethiopia

Cluster 3 Chile Oman Kazakhstan Panama Kuwait Colombia Argentina Morocco Peru Philippines Ecuador Dominican Republic Guatemala Algeria Bangladesh

The last step of our analysis consisted of choosing other variables to account for the degree of digitalization of each country, and checking whether our results are robust or not. To that end, we first turn to the Digix 2020 index from BBVA research and regress our GDP variable on the same variables as Model 1, except for the fact that we switched EDI for the Digix index. We then did the same with the World Bank Digital Adoption Index (DAI) from 2016. The results are below:

8

Table 2 : Linear regression of GDP growth (Q3 2020, Yoy) on centered mentioned variables

Digix DAI Stringency Share of services Public Deficit Widening

Regression 1 0.10**
(0.03)
-0.08
(0.06)
-0.14*
(0.07)
-0.36*
(0.15)

Regression 2
0.18** (0.05) -0.13
(0.07)
-0.36***
(0.09)
-0.48*
(0.19)

We find a consistent result with our previous analysis: our first proxy for our digitalization index, Digix, has a positive and significant influence on GDP behavior during the crisis. This time, an additional point of Digix translates to a +0.1 pp GDP resilience in y/y growth terms, which is less than the effect estimated for EDI, but of the same magnitude, and coherent with our previous values. The same can be said for the World Bank DAI and both coefficients are 99% significant. One can also note that the different degrees of significance for other variables are lessened compared to Table 1, which indicates that EDI provides a better understanding of the relation between GDP resilience and digitalization.

APPENDIX 3 : DIFFERENT MEASURES OF CENTRALITY

Different measures of state centrality were tested for this analysis. We used several variables from the Economist Intelligence Unit, such as social unrest, policy implementation and institutional effectiveness. The main downside of those variables was that they reduced our sample to 46 countries (versus 78 in Table 1). Nevertheless, here are the estimated coefficients for each regression:

Table 3 : Linear regression of GDP growth (Q3 2020, Yoy) on centered mentioned variables

EDI Stringency Share of services Public Deficit Widening Institutional Effectiveness Policy Implementation Social Unrest

Regression 1 0.28**
(0.08)
-0.09
(0.08)
-0.22*
(0.10)
-0.51*
(0.18)
-0.72
(0.50)

Regression 2 0.37***
(0.08)
-0.08
(0.06)
-0.26*
(0.10)
-0.66**
(0.18)
-3.00
(1.00)

Regression 3 0.20*
(0.08)
-0.06
(0.08)
-0.17*
(0.10)
-0.48*
(0.19)
0.16
(0.80)

Overall, all coefficients are less significant, which may be an effect of the reduction of sample size. All coefficients for the EDI variable are at least 95% significant. Additional measures of state centrality were never significant.

9

These assessments are, as always, subject to the disclaimer provided below. FORWARD-LOOKING STATEMENTS The statements contained herein may include prospects, statements of future expectations and other forward -looking statements that are based on management's current views and assumptions and involve known and unknown risks and uncertainties. Actual results, performance or events may differ materially from those expressed or implied in such forward-looking statements. Such deviations may arise due to, without limitation, (i) changes of the general economic conditions and competitive situation, particularly in the Allianz Group's core business and core markets, (ii) performance of financial markets (particularly market volatility, liquidity and credit events), (iii) frequency and severity of insured loss events, including from natural catastrophes, and the development of loss expenses, (iv) mortality and morbidity levels and trends, (v) persistency levels, (vi) particularly in the banking business, the extent of credit defaults, (vii) interest rate levels, (vi ii) currency exchange rates including the EUR/USD exchange rate, (ix) changes in laws and regulations, including tax regulations, (x) the impact of acquisitions, including related integration issues, and reorganization measures, and (xi) general competitive factors, in each case on a local, regional, national and/or global basis. Many of these factors may be more likely to occur, or more pronounced, as a result of terrorist act ivities and their consequences. NO DUTY TO UPDATE The company assumes no obligation to update any information or forward -looking statement contained herein, save for any information required to be disclosed by law.
10
CountriesDeficitVariablesDigitalizationEdi