Printer Friendly

Cross-Border Data Flows and Growth in Europe.

Abstract: From goods and people to finance and data flows, this article documents the changing nature of cross-border flows, with the relative rise of data flows versus traditional trade flows of goods and services in Europe. Within this mix change, we also report a few new important findings. First, using principal component techniques to decompose the mix of cross-border flows, we discover four factors explaining 90% of the mix variation in flows. The largest one correlates positively with all flows and can thus be interpreted as a measure of a country contribution to global flows. The component with the smallest capture of cross-border flows variance exhibits a contrasting set of correlates (positive with traditional trade, but negative with data flows) and can be interpreted as a measure of traditional, outside of digital, cross-border flows. Second, using a dynamic panel of the 28 European countries in the last 25 years, three out of the four principal components correlate with a country output growth, with a positive elasticity of cross-border flows, especially focus on data flows, to output growth. Third, cross-border data flows are the largest contributor to output growth given its large recent increase, within a material contribution of up to 0.5- to 1 point of growth for European countries.

Key words: Globalisation, economic growth, trade, data flows, principal components, error-correction model

The traditional "markers" of globalization, that is, cross-border flows of traded goods, services, and finance, were on a strong upward trajectory for two decades before the last recession.

From 1985 to 2007, the world's trade in goods grew roughly twice as fast as global GDP (McKinsey Global Institute, 2016). Since then, traditional global trade flows seem to have run out of steam: post-recession, growth has flattened, leading some people to conclude that globalization is taking a pause. For instance, using last official WTO statistics available in 2016, trade growth was just 1.3% with trade of capital goods being particularly weak, as investment spending slumped in the United States and as China was rebalancing its economy toward consumption and services, reducing import demand (1).

However, globalization has not gone into reverse; in fact, the global world, has been morphing to digital, growing more connected than before. This leads to the thesis of the "great convergence", as highlighted recently by Richard BALDWIN'S eponym book (2016), whereby ICT technologies are claimed to be changing the nature of globalization. The DHL's 2016 report on global connectedness suggests that the world was about 8% more connected in 2015 than in 2005 (2). A recent study led the McKinsey Global Institute reports on the growing importance of international data flows, pinpointing that cross-border data volume has grown by 45 times within the same time frame (MANYIKA et al., 2016). BUGHIN & LUND (2017) further add that: "by 2016, an estimated 211 terabits of data--the equivalent of 8,500 entire Wikipedias--were flowing across borders every second".

Such a large data growth can be traced to richer and richer content travelling cross-border. But in large part, the underlying driver of this growth rate is a growing intensity of Internet traffic communications and transactions. Exchange of data can also take multiple forms; it can happen "outside markets" as growing intra-firm exchanges. For example, in order to better optimize global operations, Rio Tinto, transmits data continuously from its mines, processing plants, and vehicle fleet to "excellence centers" in Australia, where analysts monitor operations in real time and head off production delays. Energy giants BP and Shell are using Internet of Things technology to explore reserves and track well production around the world. Exchange of cross-border data is also visible through markets, with nearly half of the world's traded services being already digitized (MANYIKA et al., 2016). Approximately 9% of goods trade was performed digitally by 2012 in Europe (GOMEZ-HERRERA et al., 2014). One critical channel for cross-border data exchanges relates to global platforms. Within Facebook, Twitter, Linkedln, and WeChat, close to one billion people have at least one international connection on social media, after adjusting for overlap between users of these platforms (DOW et al., 2013). Marketplaces such as Alibaba and Amazon host millions of small merchants, of which a large part is from all over the world. An E-bay study reports, that across 18 countries studied, anywhere from 88 to 100 percent of eBay sellers are exporters, while only one out of ten small firms export through physical channels (LENDLE, 2012).

In this ever diversified type of flows, together with a high rise of data flows, one natural question is how those mix changes affect a country growth path. Trade theory has long debated whether cross-border flows are good for economic development. The rise of data flows provides "another ammunition" for how they could help a country grow further. For example, the reduced barriers of distance for digital commerce combined with online platforms allowing SMBs to more easily find export markets to sell should stimulate growth; likewise, the increased amount of information flows internationally should help allocate resources better on a global scale, or ensure that global advantages are quickly built up for global corporations spreading their new innovations and ideas.

In this article, we quickly document the recent evolution in cross-border flows (Section 2); then, we assess how those flows are correlated with GDP growth path (Section 3). We focus on Euro-28 countries in the last 25 years, from 1990-2015, as a "test bed" case. Europe is already welcoming large internet users and is the cross-country block with the largest portion of intratrad: about 61% by 2014, versus 42% for Nafta, and 24% for Asia (see MANYIKA et al., 2016). Further, the timeframe of analysis includes both the rise of a common Currency Union, as well as the boom of internet connectivity since end of the nineties.

Three new findings emerge from the analyses.

1. As all cross-border flows are inter-related, we use a principal component analysis to derive orthogonal measures of flows. We find four primary components explaining up to 90% of the variance in all flows. The first component is correlated positively to all flows and can be interpreted as a measure of a country's participation to world economy. All the other three factors are more clustered to some flows, with the smallest component being negatively linked to data flows, but positively linked to all traditional flows of goods and services exchange. This component provides a good benchmark of traditional international trade which has been running out of steam lately.

2. Given trends in flows and GDP, we resort to error correction model estimation techniques as a way to appropriately assess how cross-border flows drive GDP growth. We find that flows contribute significantly to growth in both short-, and long- term.

3. Finally, using the weights in the first principal component, we find that the largest flow contributor to GDP has become cross-border data flow in recent years. Depending on country, data flows contribute to close to 0.5 point in the short term and up to one point of GDP growth after a few years, a large effect indeed, in line with the new globalization era of great convergence (BALDWIN, 2016).

* The Evolution Pattern of Cross-Border Flows

Various cross-border flows prevail as an indication of how a country has been opened to international exchange. Traditional flows include trade of goods and services, Foreign Direct Investment (FDI), as well as financial flows, including different forms of exchange, such as monetary, equity and bonds value exchanges. People flows, or migration, have come back in fashion recently with the recent rise of emigration from mostly Muslim countries to Europe as highlighted by a recent McKinsey Global Institute report (WOETZEL et al., 2016). Finally, data flows have become a major focus related to the regulation of free flows (see MANDEL, 2013).

The recent dynamics of global flows

To date, there is clear hierarchy established regarding the intensity of cross-border flows. As of 2015, our estimates, drawing on MANYIKA et al. (2016), suggest that the most internationally-oriented flows remain trade flows (about 30% of total traded revenue worldwide), while the least is migration with less than 1% of people flow happening cross-border per year. We estimate that 12% of data flows are cross-border, and 8% of investments are foreign-based.

Since the 2007 recession, the intensity of trade to GDP has declined. Among others growth of trade stalled in more developed countries. This has led to a relative rise in the proportion of trade flows coming from other countries than developed countries. While developed economies used to dominate global trade, eg 54 percent of all goods trade in 1990, these flows shrank to stand for only 26 percent by 2015.

Similar to goods and services, cross-border flows of finance, including foreign direct investment purchases of foreign bonds and equities, and cross-border lending and deposits, have yet to reach their 2007 peak. While increasing from $470 billion in 1980 to a peak of $12 trillion by 2007, the financial crisis triggered a sharp retreat. Banks had to suffer a long period of deleveraging and retrenchment, and cross-border lending fell by some 96 percent in 2012, most notably within Europe. FDI has held up better than other types of financial flows and now accounts for about half of financial flows.

The portion of people living outside their home country has remained remarkably steady at 2.7 percent since 1980. Forced migration can change this pattern at time of wars, as it happens recently with major instability in Middle-East. A higher share of migrants is nevertheless moving from emerging to developed economies--(29 percent of migrants in 1990 compared with 47 percent in 2015, see WOETZEL et al., 2016). Students enrolling in foreign universities have grown at 3.4 percent and 4.8 percent per annum, respectively, between 2002 and 2010.

In contrast to the traditional flows, data flows have been growing steadily across the globe. Since the first emergence of commercial dial up internet connections roughly 25 years ago, a fast and growing portion of people, companies and objects has started to be connected and to share ideas, transact, make social connections, or exchange real time information. In BUGHIN & LUND (2017), we report a measure of data flows in bits per year, passing either though the public internet or through digital private networks. For this computation, we used the most extensive private data sets captured by Tele-geography, a private firm which tracks capacity and use of the global network of submarine optical fibre cables. We found that, since 2005, the volume of data flows, measured in terabits per second, has multiplied by a material factor of 45 in a decade, to reach an estimated 400 terabits per second by the end of 2016.

Data flows unique features

Besides its sheer growth, a few features are worth pinpointing regarding data flows:

a) The cross-border data intensity depends on the type of flows. For example, the digital share is significantly larger when the underlying product is digitized. For physical products, the amount of cross-border digital commerce is typically less than 10% still, but it grows to more than 30% for international communication worldwide in 2016. Among news and entertainment companies, close to 80% of bits traffic originates internationally for the Financial Times, 60% for BBC, and up to 50% for NewsFeed and Netflix (BUGHIN & LUND, 2017).

b) Cross-border data flows remain concentrated to date. Data flow intensity is large in developed countries, reflecting both much larger connectivity adoption and lower cost of dense international routes. In fact, even with populations getting largely converted, cross-border data of European countries have grown by 60% between 2013 and 2008, or a 9-fold intensity of cross-border data in 65 years. Further, there is a catching up process happening: small and more isolated countries such as Cyprus and Malta in Europe have built up cords border data a 20-fold, or three times the average.

c) This surge of cross-border data not only constitutes a massive flow in and of itself but is also transforming the other types of flows. Consider for example how the internet is changing the patterns of trade commerce. While trade intensity tends to concentrate between close neighbouring countries (a phenomenon dubbed "gravity" by trade economists), patterns of e-commerce exchanges seem to spread wider, connecting countries from much further apart. FREUND & WEINHOLD (2004) used a gravity model to examine the effect of the Internet on trade among 56 countries. They found an increasing and significant impact, already from 1997 to 1999. TIMMIS (2012) confirmed the same effect of internet adoption on trade for OECD countries for the period 1990-2010. FINK et al. (2005) further found that, along with lowering the fixed cost, the Internet tends to reduce the variable cost of trade and thereby augments the trade volume. More recently, GOMEZ-HERRERA et al. (2014) demonstrated that the depressive effect of distance on cross-border trade decreases by half when products are transacted via online platforms.

d) Using European countries, we plotted country GDP growth against cross-border trade of goods on one hand, and against growth in cross-border data, on the other hand. The correlation pattern is the same for each year since 2007, and in particular, taking the yearly average over 2000-2007, the correlation is negative for tradtional trade flows (r= -0.32), but positive for data flows (r= 0.14). As illustrative at best, those correlations make the point of a possible different set of dynamics for traditional and new, data-driven, flows. Let alone, they may be suggestive of a possible stimulating effect between data flows and economic growth, in a time of stalling intensity of traditional trade.

Measuring country connectedness

Country connectedness

We are not interested in cross-border flow intensity per country per se, but more on how each country contributes its share to cross-border flows. Flow intensity leaves the world's largest trading and exchanging countries, like the United States, China or Germany, looking comparatively closed because of the large size of their domestic markets. Conversely, smaller countries such as Luxembourg inevitably have larger flows of goods, services, and finance compared with the size of their economies.

To account properly for a country's significance in world flows, we, therefore, also need to assess its global share of trade and exchange. Hereafter, we use the measure proposed by SQUALLI & WILSON (2006) to derive the "connectedness" of a country by multiplying flow intensity by its share of global flows. Taking both measures into account corrects the tendency for small/large countries to rank high/low on cross-border flows.

Flows clustering

As most types of flows are likely interrelated, we are also interested in better qualifying flows, building more clear-cut and orthogonal vectors of flow types per country. One side benefit of building orthogonal flows is also that we can use those new constructs to estimate stable regression correlates between output growth and cross-border flows.

Hereafter, we report on performing a principal component analysis on cross-borders flows. The analysis is concerned with 9 types of flows, splitting trade of goods and services, into trade of goods, trade by service types, FDI flows, people flows, travel flows and bit data flows. The analysis clearly emphasizes that 4 factors synthesizes 90% of the variance in flows, and the first component already explains 2/3 of the variance in total flow connectedness (Figure 1).
Figure 1 - Principal component analysis of cross-border flows
(Varimax rotation)

Principal Component Analysis on the Connectedness Index

Component  Eigenvalue  Difference  Proportion   Cumulative

Comp1      5,98        5,18        66,4%        66,4%
Comp2      0,80        0,05        8,9%         75,3%
Comp3      0,75        0,26        8,3%         83,6%
Comp4      0,49        0,17        5,5%         89,1%
Comp5      0,32        0,03        3,6%         92,7%
Comp6      0,29        0,08        3,2%         95,9%
Comp7      0,21        0,08        2,4%         98,3%
Comp8      0,14        0,12        1,5%         99,8%
Comp9      0,02                    0,2%        100,0%


Checking further on those 4 principal components (Figure 2) each of them can be given a clear explanation. Each flow contributes positively and in rough proportion to the first principal component, and can thus be interpreted as a measure of how a country contributes to the global world network of global flows.

The second component is positively correlated with, and over-indexed on, FDI; similarly for the third component over-indexed on people flows. The second (respectively, third) component can be thus be interpreted as an indicator of a country contribution to FDI flows (respectively, as an indicator of a country contribution to people flows).

The fourth principal component is shown to be most positively correlated to traditional trade of goods and service goods- but negatively linked to both data flows and knowledge intensive services. It thus captures more or less the country bias to traditional trade rather than new data trade.
Figure 2 - Flow weights to principal component constructs

Principal Component Analysis on the Connectedness Index: contribution
of index (top 4 components)

Variables                                         components
                                              Comp1       Comp2

In(Connectedness Index for goods trade)       36,3%        -3,5%
In(Connectedness Index for service trade)     39,2%       -15,3%
In(Connectedness Index
for service trade - knowledge intensive)      35,7%        -3,1%
In(Connectedness Index for service trade      34,6%       -11,2%
- labor intensive)
In(Connectedness Index for
service trade - capital intensive)            34,1%       -25,8%
In(Connectedness Index for FDI flow)          20,8%        94,5%
In(Connectedness Index for migrants)          23,8%         3,8%
In(Connectedness Index for travelers)         37,3%        -0,8%
In(Connectedness Index for internet traffic)  33,4%         2,3%
proportion var explained                      66,4%         8,9%

Variables                                       components
                                              Comp3   Comp4

In(Connectedness Index for goods trade)        -6,5%   20,4%
In(Connectedness Index for service trade)     -10,8%    8,6%
In(Connectedness Index
for service trade - knowledge intensive)      -19,9%  -31,2%
In(Connectedness Index for service trade       12,0%   45,5%
- labor intensive)
In(Connectedness Index for
service trade - capital intensive)            -20,4%   35,6%
In(Connectedness Index for FDI flow)           -9,9%   18,2%
In(Connectedness Index for migrants)           91,4%  -12,7%
In(Connectedness Index for travelers)           8,3%  -16,1%
In(Connectedness Index for internet traffic)  -19,0%  -66,7%
proportion var explained                        8,3%    5,5%


* Cross-Border Flows and Economic Growth

In this section, we are testing how those different flows contribute to economies, in particular how they are linked to economic growth. As our analysis demonstrates, we are finding that they are, but emphasizing the importance of data flows, as a new growth engine.

Literature review

Many recent empirical studies which have been analyzing the link between cross-border flows and growth rely on the new growth theory that suggests that international economic integration, and cross-border spillovers from innovation are important channels to the growth process.

If the consensus has been established that the relationship between cross-border flows and economic growth is positive on theoretical ground, the empirical evidence, while often looking at one peculiar flow at a time, has not been always as clear-cut. Regarding financial flows, KRUGMAN (1993) had argued both ways on how financial flows improve economic growth. Regarding FDI, BORENSZTEIN et al. (1998) investigated and found a positive effect of FDI on growth, but their work also highlighted that it runs mostly through human capital complementarity with FDI.

Regarding goods and service trade flows, a seminal study by SACHS et al. (1995), which argued that trade flows of goods and services promote growth, has been challenged later by RODRIGUEZ & RODRIK (2000), on the ground of absence of control variables in the regressions, likely introducing bias in the estimated relationship between flows and growth.

The question whether migration is linked to growth has featured prominently in the academic literature recently, due to a recent growth of migration flows. Besides the consensus around the beneficial impact of immigration on the economic development of the host country (ORTEGA & PERI, 2014), the issue of migration cannot be easily separated from the one on cultural diversity; and so far, it has been difficult to establish a robust direction of the effect of cultural diversity on economic growth.

Finally, analyses on how data flows build growth remain scarce, as the phenomenon of data explosion is rather new, and data evidence is difficult to collect. One exception in the series of recent work by the McKinsey Global Institute, see MANYIKA et al. (2014, 2016) using private data collected by Telegeography of cross-border bits flowing through the public Internet and private networks. Those works are able to find a positive effect on growth from data flows, among others. The estimates imply that cross-border data flows have contributed to world GDP by about 3 percent, adding $2.8 trillion to global GDP by 2015 (BUGHIN & LUND, 2017).

Cross-border flows and growth: new estimates for Europe

Hereafter, we develop a pooled time series- panel econometric analysis of the effects of all flows on country growth for the 28 countries of Europe, during 1990 to 2015. The focus on Europe is motivated by two features. Europe concentrates a large part of data flows to date, and furthermore, is the most integrated block, with intra trade weighting m2/3 of all cross-border flows of goods and services.

The additional peculiarity of our analysis is manifold. First, we are not focusing on one peculiar cross-border flow, but on all flows as in MANYIKA et al. (2016). Second, given that flows are linked to each other, creating risk of large multicollinearity in estimates, we leverage the principal component vectors as regressors. Third, our econometric specification is one of an error-correction mechanism, which reflects the fact that variables at hand are integrated of order one.

Empirical strategy

Our baseline empirical model is the standard accumulation growth model. We thus postulate the following form (1):

[g.sub.i,t] = [alpha] + [eta][f.sub.i+] +[phi] x t+ [beta]. [PCF'.sub.i,t] + [chi]' x [C'.sub.l,t] + [e.sub.l,t] (1)

where:

- [g.sub.i,t] is the annual percentage growth rate of the (PPP Converted) per capita GDP at 2005 constant prices in country i over a specific time, t;

- [PCF'.sub.i] is the vector of principal components;

- C' is a vector of exogenous control variables that are known to affect growth rate outside of cross-border flows. The vector C includes the country stock of capital K, as well as the amount of labour L. We also have included the share of highly educated people as a measure of human capital, HC, enhancing the labour elasticity of growth

- Further, [alpha], [beta], [eta], [phi], X are all coefficients to be estimated, while f and t are respectively country and time fixed effects, [epsilon] is an error term. We expect that X>0, and we test the main hypothesis that [beta]>0. The equation (1) is furthermore transformed in log, so that coefficients are elasticities.

An important issue with model (1) is the non-stationary of variables, while we may also want to test for the likely endogeneity of flows. Regarding the latter, our strategy consists in exploiting the dynamic nature of our panel, and using lagged values of flow as instruments (ARELLANO & BOVER, 1995). Regarding the first issue, we have tested our variables for unit root and confirm that we cannot reject that variables are trending (3)--hence, we resort to estimating an error correction model of the form (2), as a proper representation of the dynamics of the variables (ENGLE & GRANGER, 1987):

[DELTA][g.sub.i,t] = [lambda] + [mu]. [DELTA][PCF'.sub.i,t] + v. [DELTA][C'.sub.l,t] + [delta]. [g.sub.i,t-r] + [theta]. [ER..sub..[IOTA]]+ [e'.sub.l,t] (2)

- Which is equation (1) in first difference [DELTA] including the residual of the estimated growth [g.sub.it]* from the regression in level from equation (1), that is: ERit = [g.sub.it]*-[g.sub.it] (4).

- We expect that [theta] < 0, demonstrating sign of convergence to long-run and that the error-correction model is well specified (5).

- We added as well lagged output as regressor, as a result of the theory of growth stipulating a catch up process,

- thus we expect 0 < 0.

Final results and interpretation

We report final results of the system of equations (1)-(2) in the next three tables. For each table, the instruments are one year lagged flow for long-term model, while we use lagged three and more as the panel proceeds in time for the short-term model. The lag growth variable we use is 3 years, which means our model is estimated from 1995 to 2015. For simplicity only the regressors with statistical significance at 10% are reported. When statistically significant at 10%, the level of risk is also reported below the estimated elasticities. We do not report country effects as nearly all were not significant.

The first model (Figure 3) focuses on traditional flows; that is international goods and services. We thus omit other types of flows such as people flows (and mostly data flows). The second model (Figure 4) focuses only on data flows, and neglects all other flows. Both models may suffer from omitted flow variables.

The last model includes all the four principal components of flows (Figure 5). Looking at the (adjusted) R-square values as a measure of fit, they are very evidently large for equation (1) as is typical when unit roots are present in the dynamics of the variables the model. The fit from equation (2) is a more relevant test, and R-square values are confirmed to be the largest both within and between observations for the most comprehensive model using all principal components.

In general, whatever the focus, the estimates are in line with expectations. First, labour elasticity with respect to output is roughly 60-65% both for short term and long-term; the capital elasticity is relatively large in short term, but is reduced to about 35% in long term. Summing the two elasticities, we find roughly constant to scale, at about the value of 1. Second, the lagged growth has a negative elasticity, meaning that past growth reduces the chance of higher growth in future period. Third, the error-correction residual is always negative, demonstrating again that short term deviations revert to long-term values, as a symptom of an equilibrium structural relationship between regressors (thus, cross-border flows) and country growth.

We now turn to the relationship between flows and growth. In the first table, goods and services exhibit long-term positive elasticities, as expected from the tenants of traditional globalization. However, three remarks are in order. The first remark is that only knowledge services flows appear to have significant short as well as long term effects; those type of flows are also most correlated to (unobserved) data flows as witnessed in our fourth principal component. The second remark is that internet and ICT technologies are omitted variables in this equation, and we know from the literature review that they are usually reinforcing the apparent relationship between traditional flows and growth. This leads to a possible estimation bias in the trade flow effect on growth as reported in figure 3. Finally, the intensity of trade as measured by traditional flows has been nearly flat in recent years, -on average growing by one point of percentage a year. This means that at the current elasticity of goods and services traded, the contribution to growth of flows has been very low, barely 0.1 point of percentage for the average country in Europe.
Figure 3 - Error-correction model of growth and traditional trade flows

                     Long
                     Term                              short term
Variable             equation (1)  Variable            equation (2)

Cross-border flows:
Goods                 0,1288                           n.a.
                           0
knowledge services    0,0756       knowledge services  0,01097
                           0                            0,0939
labor services         0,048
                      0,0002
Control variables:
K                     0,3765       [DELTA]K             0,7239
L                          0       [DELTA]L                  0
                      0,6214                            0,6283
                           0       lagged g                  0
                                                        -0,051
                                   ER                   0,0422
                                                       -0,3399
Time effects                                                 0
T_2003
                      -0,069
T_2004                     0
                     -0,0368
T_2007                     0
                      0,0253
T_2008                0,0001       T_2008
                      0,0141                           -0,0273
T_2009                0,0267       T_2009                    0
                     -0,0403                           -0,0554
Constant                   0       constant                  0
                     -0,9978                            0,5183
r2 within              0,007       r2_w                 0,0399
r2 between                87%      r2_b                     80%
                          97%                               34%


In figure 4, we concentrate on data flows only. We notice that both short-term and long-term data elasticities are positive. This provides some good ammunition of the evidence of a relationship between cross-border data exposure and country growth. We note an estimated coefficient for Europe in the range of the elasticities found in MANYIKA et al. (2016), but for a global sample of more than 120 countries. The speed of adjustment is rather rapid, in less than 5 years. If data flows are large indeed, their intensity (cross-border versus within country) remains low, and as discussed, traditional trade flows are likely an omitted variable in figure 4.
Figure 4 - Error-correction model of growth and data flows

            Long                    short
            Term                    term
                                    equation
Variable    equation (1)  Variable  (2)

Data         0,1134                   0,0522
              0,002                    0,074
K            0,3821       [DELTA]K    0,7899
                  0                        0
L            0,6045       11          0,6211
                  0                        0
                          lagged g    -0,067
                                      0,0422
                          ER         -0,2235
                                      0,0021
T_2003       -0,061
                  0
T_2004      -0,0428
                  0
T_2007       0,0271
                  0
T_2008      0,02231       T_2008    -0,02532
             0,0267                        0
T_2009      -0,0421       T_2009     -0,0456
                  0                        0
constant      1,045       constant    0,1272
              0,008                   0,0026
r2 within        88%      r2_w            82%
r2 between       93%      r2_b            29%


We solve this in figure 5, which makes full use of the four principal components. While the second one, as a measure of FDI contribution to flows, does not appear to be significant, the three other components are.

The first component exhibits the largest elasticity of the three estimates. It doubles from short to long-term, while adjustment to long-term takes just below four years (inverse of the coefficient of the error correction residual; 1/27%= 3.7 years). The third component relates to migration flows and shows a reversal of elasticity, from negative in the short -term to positive in longer term. Those dynamics are roughly in line with WOETZEL et al. (2016), but elasticities appear relatively low versus other works (iNANC-TUNCER, 2016). The elasticity linked to the fourth principal component is negative both short and long -term. Remember that the weights of flows into this component are mostly positive with respect to traditional flows and negative with respect to data flows. Otherwise stated, this component is a measure as to how a country focuses on traditional versus new types of flows. The negative elasticity suggests that this reliance on traditional flows is possibly limiting a country growth path.

We finally use the principal component elasticities to derive the full effects of cross -border data growth into each of the European economies, as reported in figure 6 for long term effects.

As intensity of data flows has increased significantly in recent year at double digits, the estimated effect is large, in the long-term up to 1 point of growth. In the short-term, the effect is about half and builds up to its long-term in about 4 years. Those estimated effects on growth are rather significant, with large contribution for Ireland and much less for Luxrembourg (depending on dynamics of data globalization) and illustrate the changing nature of globalization, as among others highlighted by authors such as BALDWIN (2016) and BUGHIN & LUND (2017).
Figure 5 - Error-correction model of growth and principal components of
all flows

            Long                      short
            Term                      term
                                      equation
Variable    equation (1)  Variable    (2)

pc1           0,094                    0,0522
                  0                     0,074
pc3           0,021
              0,013
pc4          -0,021
              0,092
K            0,3414       [DELTA]K     0,8193
                  0                         0
L            0,6704       [DELTA]L     0,6193
                  0                         0
                          lagged g    -0,0947
                                            0
                          ER          -0,2781
                                            0
T_2003      -0,0524
                  0
T_2004      -0,0233
             0,0061
T_2007       0,0276
             0,0002
T_2008       0,0243       T_2008      -0,0201
             0,0011                         0
T_2009      -0,0447       T_2009      -0,0566
                  0                         0
T_2010      -0,0221       T_2010       0,0214
             0,0023                         0
Constant      1,045       Constant     0,1272
              0,008                    0,0026
r2 within        89%      r2_witihn        87%
r2 between       98%      r2_between       36%

Figure 6 - Estimated long-term effects of data flows on European
country growth

Contribution to nominal growth, 2013-2012, Euro 28

Luxembourg      0.40%
Cyprus          0.59%
Denmark         0.62%
Italy           0.64%
Greece          0.79%
Romania         0.82%
Slovenia        0.90%
Spain           0.96%
Hungary         1.05%
Lithuania       1.19%
United Kingdom  1.19%
Poland          1.20%
France          1.25%
Slovak
Republic        1.29%
Malta           1.29%
Croatia         1.31%
Germany         1.32%
Portugal        1.32%
Netherlands     1.46%
Bulgaria        1.54%
Sweden          1.62%
Latvia          1.75%
Finland         1.90%
Czech Republic  1.94%
Estonia         2.11%
Ireland         2.78%


* Conclusions

The material increase of cross-border data flows has been noticed by multiple trade economists (e.g. BALDWIN, 2016), and their unique nature should have both different and material impact on globalization and country ability to grow (MANYIKA et al., 2016, or MANDEL, 2016). In this article, we document this shifting nature in flows, and provide robust evidence as to the effects of cross-border flows on economic growth.

In our more complete model, which uses a principal component of all types of flows, we find that a country contribution to all cross-border flows, especially focused towards data flows has significant effect on a country's ability to grow. Given the recent explosion in data, the potential in the long-term of cross-border data flows can be rather significant in Europe, up to one point of growth.

Recognizing the importance of data flows, the European Commission has outlined a Digital Single Market agenda focused on bolstering consumer protections, improving logistics, and simplifying VAT administration. It also includes harmonized regulations on privacy and cybersecurity. But policy makers across the continent need a real sense of urgency on these issues: today only 15 percent of EU consumers purchase online from other EU countries, and only 7 percent of Europe's small and medium-sized businesses sell cross-border. Europe is also challenged in its ability to be at the core of cross-border data networks (6); at the exception of The Netherlands adding a virtual data leg to its harbour infrastructure, one witnesses a slow drop-off especially on Northern countries which used to be most opened to data flows. Notably, Scandinavia and Belgium are slowly migrating to the periphery of those flows.

Not as unusual, China is catching up. For instance, more than 18 percent of China's trade already takes place on digital platforms--approximately double the share in Europe. Besides Alibaba's huge hold on B2B, multiple online B2B companies have been set up in China for global extension in many verticals such as steel and textiles. Cross-border data flows are an important feature of the new globalization and every player should take actions.

References

ARELLANO, M. & BOVER, O. (1995): "Another look at the instrumental variable estimation of error-components models", Journal of Econometrics, 29-51.

BALDWIN, R. (2016): The great convergence: information technology and the new globalization, Harvard University Press, Cambridge.

BARRO, R. & SALA-i-MARTIN, X. (1992): "Convergence," Journal of Political Economy, 100, no. 2, pp 223-251.

BISWAS, T. & KENNEDY, P. L. (2016): "The Effect of the Internet on Bilateral Trade", in 2016 Annual Meeting, February 6-9, San Antonio, Texas (No. 229994), Southern Agricultural Economics Association.

BORENSZTEIN, E., De GREGORIO, J. & LEE, J (1998): "How does foreign direct investment affect economic growth?", Journal of International Economics 45: 115-135.

BUGHIN, J., LUND, S. & MANYIKA, J. (2015): "Harnessing the shifting power of global flows", McKinsey Quarterly.

BUGHIN, J. & LUND, S. (2017): The ascendancy of international data flows, in Voix Europa, 09 January, http://voxeu.org/article/ascendancy-international-data-flows.

CLARKE, G. R & WALLSTEN, S. J. (2006): "Has the internet increased trade? Developed and developing country evidence". Economic Inquiry 44.3, pp. 465-484.

DOW, A., ADAMIC, L & FRIGGERI, A. (2013): "The anatomy of large Facebook cascades," presented at Seventh International Conference on Weblogs and Social Media in Cambridge, Massachusetts, July 8-11.

ENGLE, R. & GRANGER, W. (1987): "Error Correction: Representation, Estimation, and Testing", Econometrica, Vol. 55, No. 2. Pp 1-17.

FINK, C, MATTOO, A. & NEAGU, I. (2005): "Assessing the impact of communication costs on international trade", Journal of International Economics, 67.2, pp. 428-445.

FRANKEL, J. A. (2008): The world trading system and implications of external opening, in: N. Serra, J.E. Stiglitz (eds.), The Washington consensus reconsidered: towards a new global governance, Oxford, University Press, Oxford.

FREUND, C. & WEINHOLD, D. (2004): "The effect of the Internet on international trade", Journal of International E conomics, 62(1), pp. 171-189.

GOMEZ-HERRERA, E., MARTENS, B. & TURLEA, G (2014): "The drivers and impediments for cross-border e-commerce in the EU", Information Economics and Policy, 28, 83-96.

KRUGMAN, P. (1993): International Finance and Economic Development, in: Giovannini A, editor, Finance and Development: Issues and Experience, Cambridge University Press. 11-24.

INANC-TUNCER, O. (2016): "The Labor Market Effects of Immigration", Handbook of Research on Unemployment and Labor Market Sustainability in the Era of Globalization: 388.

LENDLE, A., OLARREAGA, M., SCHROPP, S. & VEZINA, P.-L. (2012): "There goes gravity: how eBay reduces trade costs", CEPR discussion paper, London.

MANDEL, M. (2013): "Data, Trade and Growth", TPRC 412: The 41st Research Conference on Communication, Information and Internet Policy, The Progressive Policy Institute, March. http://ssrn.com/abstract=2241302.

MANYIKA, J., BUGHIN, J., LUND, S., NOTTEBOHM, O, POULTER, D., JAUCH, S. & RAMASWAMY, S. (2014): "Global Flows in a Digital Age: How Trade, Finance People, and Data Connect the World Economy", McKinsey Global Institute, April.

MANYIKA, J., LUND, S., BUGHIN, J., WOETZEL, J., STAMENOV, K & DHINGRA, D. (2016): "Digital globalisation: the new era of global flows", McKinsey Global Institute, February.

ORTEGA, F. & PERI, G (2014): "Openness and income: The roles of trade and migration", Journal of International Economics 92 (2014), 231-251.

RODRIGUEZ, F. & RODRIK, D. (2000): "Trade policy and economic growth: a skeptic's guide to the cross-national evidence", NBER Macroeconomics Annual 15: 261-325.

SACHS, J., WARNER, A., ASLUND, A., FISCHER, S. (1995): "Economic reform and the process of global integration", Brookings Papers on Economic Activity 1995: 1-118.

SEO, H. & THORSON, S. J. (2012): "Networks of networks: Changing patterns in country bandwidth and centrality in global information infrastructure, 2002-2010," Journal of Communication, volume 62, number 2, April.

SQUALLI, J. & WILSON, K. (2006): "A new approach to measuring trade openness", Economic Policy Research Unit, Zayed University, Dubai, working paper number 06-07

WOETZEL, J, MADGAVHAR, A., RIFAI, K., MATTERN, F., BUGHIN, J., MANYIKA, J., ELMASY, T., Di LODOVICO, A. & HASYAGAR, A. (2016): "People on the move: global migration's impact and opportunity", McKinsey Global Institute, December.

Jacques BUGHIN

Director, McKinsey Global Institute and senior partner, McKinsey & Company, Fellow University of Brussels & KUL, Leuven, Belgium

(1) The interested should refer to https://www.wto.org/english/res_e/statis_e/statis_e.htm.

(2) http://www.dhl.com/en/about_us/logistics_insights/studies_research/global_connectedness_index/global_connectedness_index.html

(3) We do not report the results here--we used typical Augmented Dickey-Fuller (ADF) unit tests.

(4) Coefficients in (2) are the short-term elasticities counterpart of equation (1). Given first difference, fixed effects cancel out in equation (2).

(5) The test is based on the ENGLE-GRANGER co-integrated test with ADF.

(6) For a social networks analysis of communication and data across countries see among others SEO and THORSON (2011)
COPYRIGHT 2017 IDATE
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Bughin, Jacques
Publication:DigiWorld Economic Journal Communications & Strategies
Date:Jul 1, 2017
Words:6558
Previous Article:Trends in Global ICT Trade.
Next Article:International ICT Trade Dynamics 2004-2014: an explorative network analysis.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters