Economic Implications of Data Regulation

4. Outcomes of the simulations

4.1 Changes in data-flow regulation can have sizeable impacts

Overall impact

Four main takeaways emerge from the simulations on data flow regulation (Figure 4.1). (A discussion of the channels of transmission can be found in Annex C.1).

• The no regulation scenario (Scenario 1) underscores the economic importance of data flow policies to build trust in economic transactions. Moving to no regulation leads to overall reductions in exports and GDP because the losses from a reduction in trust outweigh the gains from lower trade costs as a result of less regulation.

• The scenario where all regions move to the middle (Scenario 2) shows the benefits of convergence towards balanced data flow regulations in the form of regimes with either open or pre-authorized safeguards. Here benefits are sizeable, with a projected increase in global exports of 3.6% and in global GDP of 1.77%.

• The geoeconomic fragmentation scenario (Scenario 3) highlights that the economic costs of fragmentation of data flow regimes are sizeable, with global exports projected to fall by about 1.76% and global real GDP to fall by 0.94%.1

• The data doomsday scenario (Scenario 4) shows the overall importance of data flows in underpinning economic activity. A move to “data autarky” would see real global exports fall by 8.45% and real global GDP by 4.53%.

Figure 4.1. Projected change in global exports and GDP as a result of different data flow regulation scenarios

Note: This figure displays the projected change in percentage of real exports of goods and services and real GDP at the global level under the four scenarios.

Source: Simulations with the WTO Global Trade Model (version with Melitz market structure as described in Bekkers and Francois (2018[34]).

Impact by level of development

The projected changes in these scenarios do not fall equally across economies across different income groups:2

• High income economies would lose most from the no regulation scenario (GDP losses around 1.5%). This reflects losses in trust benefits from existing regulation. By contrast, low- and lower-income economies would see small benefits arising from reductions in trade costs as regulations are eased (just below 0.5% of GDP).

• The main beneficiaries of moving to the middle would be low- and middle-income economies. These would see trade costs fall and trust increase leading to increases in GDP of 5 and 4.3% respectively. High-income economies also benefit, but to a much lesser extent (increases in GDP of 0.5%), and largely as a result of positive spillover effects from changes in low- and middle-income countries.

• Losses arising from geoeconomic fragmentation would be concentrated in high and upper middle-income economies (losses of 0.9% and 1.2% in GDP). Low-income economies would experience losses of about 0.5% of GDP, while lower middle-income economies would gain (+0.2% of GDP), given that many lower-middle income economies are not part of a bloc (and would therefore not restrict their flows vis a vis others).

• Under the data flows doomsday scenario, the largest losses would be experienced by high-income economies who would face both higher trade costs and less trust leading to losses of around 5.3% in GDP. Low-income economies would generally experience smaller losses (around 2.5%), reflecting the fact that many of these economies already have either no policies (thus experiencing less trust) or restrictive policies (which already incur trade costs) however, low-middle income economies would experience sizeable shocks (close to 4 % reductions in GDP).

  To an extent, this analysis suggests that policies in low and middle-income economies that restrict data flows for digital industrial policy purposes might be counterproductive. This is because they tend to increase trade costs, but without having an offsetting positive impact on trust. For these countries, significant losses are associated with just imposing restrictions on data flows, but gains are largest from measures that allow for data to flow with conditions that help generate trust. That said, some caveats apply: the economic model employed here does not account for other channels which some argue might make restrictive digital industrial policy effective, i.e. the presence of scale economies and network externalities. The potential benefits of restrictions stemming from these characteristics would have to be assessed against the costs that such measures impose. Models of international trade with such features are not available (see Ciuriak (2017[36]) for a discussion).

Figure 4.2.  Projected change in real exports and real GDP by income group from changes in data flow regulation

Note: The figure displays the projected change in percentage of real exports of goods and services and real GDP at the global level under four scenarios: (1) no regulation; (2) moving to the middle; (3) geoeconomic fragmentation; (4) data flows doomsday for four income groups: low-income economies (less than USD1 006), lower-middle income economies (between USD1 006 and USD 3 955, higher middle-income economies (between USD 3 956 and USD 12 235), and high-income economies (more than USD12 235) employing the World Bank definition with cutoffs for 2017, the baseline year of the simulation data.

Source: Simulations with the WTO Global Trade Model (version with Melitz market structure as described in Bekkers and Francois (2018[34]).

Impact by region

There is a strong degree of heterogeneity across regions, largely depending on the current regulatory stance (Figure 4.3). In the first scenario, no regulation, as foreshadowed, regions which would see both trade costs fall and willingness to pay decrease (Categories 1 and 2) would face important reductions in both exports and GDP. This would include Europe (EUR) and North America (NAM). By contrast, under this scenario, exports and GDP for Central Asia (CEA), which comprises a number of economies with Category 3 measures, are projected to rise, reflecting reductions in trade costs with no associated changes in trust.

Figure 4.3. Projected change in real exports and real GDP by aggregate region from changes in data flow regulation

Note: The figure displays the projected change in percentage of real exports of goods and services and real GDP at the global level under four scenarios: (1) no regulation; (2) moving to the middle; (3) geoeconomic fragmentation; (4) data flows doomsday for ten aggregate regions: Central Asia (CEA); East Asia (EAS); Europe (EUR); Latin America and the Caribbean (LAC); Middle East and Northern Africa (MIN); North America (NAM); Other Asia (OAS); Oceania (OCE); Southeast Asia (SEA); Sub-Saharan Africa (SSA).

Source: Simulations with the WTO Global Trade Model (version with Melitz market structure as described in Bekkers and Francois (2018[34]).

The Asian regions (CEA, EAS, OAS and SEA), which comprise economies in categories 0 and 3, are projected to benefit in the second scenario, moving to the middle, seeing reductions in trade costs but also associated increases in trust (willingness to pay). Furthermore, under this scenario Sub-Saharan Africa (SSA), which comprises a number of economies in Categories 0 and 3, is projected to gain because the effect of increased trust/WTP as a result of the introduction of data flow regulation dominates the increase in trade costs. For the other regions, there are no changes in trade costs, so the trade and GDP effects are second order (i.e. arise through changes in what others are doing).

In the third scenario, geoeconomic fragmentation, the projected trade and GDP effects are negative for almost all regions. The effects are largest for Central Asia (CEA), East Asia (EAS) and the Middle East and Northern Africa (MIN). These regions contain most countries in the Eastern Bloc and would thus be more adversely affected than regions in the Western Bloc such as Europe (EUR) and Northern America (NAM). The latter regions tend to trade most within the Western Bloc and are thus less affected by geoeconomic fragmentation. In this scenario, the effects for regions that are in neither bloc, such as OAS (Other Asia) and Sub-Saharan Africa (SSA), are smaller, which is expected because these regions do not face cost increases vis-à-vis the other blocs (by assumption).

Finally, in the fourth scenario, data flows doomsday, the projected losses for all regions are large, with GDP losses ranging between about 6% for Europe (EUR) to 11% for Southeast Asia (SEA). For the European Union, the more modest effects reflect the fact that intra-EU trade is not subject to increases in trade costs. This is not the case, however, for trade within SEA, where trade costs are projected to rise.3

Impact on trade patterns

Regional changes are, in part, driven by changes in trade patterns (Annex C2). Four key messages emerge from the trade pattern analysis.

• Trade shifting plays an important role in the projected changes. For example, in the no regulation scenario (Scenario 1), exports from regions which previously had open or pre-authorised safeguards are projected to fall, because demand shifts to regions which previously had ad hoc authorisation (due to lower trade costs in these regions and to lower trust in regions which had pre-existing open or pre-authorised safeguards).

• Intermediate linkages are an important explanation for the projected changes in trade patterns. For example, in the no regulation scenario (Scenario 1), exports from regions without regulation to regions which formerly had ad hoc authorisation are projected to rise, because of increased production in the latter regions leading to more demand for intermediate imports.

• Bilateral trade changes in the geopolitical scenario (Scenario 3) reflect changes in trade costs. Hence, trade between the Eastern and Western Blocs falls, whereas trade within blocs and between regions outside these and the Eastern and Western Blocs rises.

• The data flows doomsday scenario (Scenario 4) sees the largest reductions in exports. This is the case for all regions. However, impacts are smallest, and sometimes even positive, for those regions that already have the most restrictive policies (Category 3). Indeed, exports of Category 3 regions to other Category 3 regions increase, largely as a result of the strong reduction in exports of all other regions as a result of growing trade costs.

4.2. Impacts of data localisation tend to be smaller, but depend strongly on the measure applied

Overall impact

When looking at the projected global exports and GDP changes in the different scenarios for data localisation measures (Figure 4.4), the following messages emerge.

• No Data Localisation scenario (Scenario A) reveals that, while global GDP is projected to increase, the impacts of removing data localisation measures are, in relative terms, smaller than those of removing data flow regulation. This is because data localisation policies generate small cost increases (according to the business questionnaire) through the suboptimal use of data management inputs which represent a small share of total costs. Minor changes in trade costs and trust arise because there are economies that have very restrictive data localisation policies with less restrictive data flow policies. Once data localisation measures are removed, these economies would see a reduction in both trade costs and an increase in trust to the level of the data flow regime. Overall, moving to no data localisation would increase exports by 0.26% and GDP by 0.18%.

• In Horizontal Storage Conditions scenario (Scenario B), global exports and GDP are projected to increase very slightly. This is because the positive impact of economies moving from restrictive data regulation to less restrictive horizontal storage conditions outweighs the negative impact of economies moving from no regulation to horizontal storage conditions (e.g. increase in data management costs). Overall, moving to horizontal storage conditions in regions would raise global exports and global GDP by respectively 0.19% and 0.04%.

• In the Sectoral Storage and Flow Prohibition scenario (Scenario C), real GDP and export losses are seen to be much larger than in previous scenarios, even if they are only implemented in a small number of sectors. This suggests that the economic costs of localisation conditions depend strongly on both the type of measure and the sectors in which these are implemented. Overall, the moving to sectoral storage and flow prohibitions is projected to decrease exports by0.95% and GDP by 0.56%.

• The Horizontal Storage and Flow Prohibition scenario (Scenario D) underscores the difference between applying storage and flow prohibitions across the board or only in a limited number of sectors. Global GDP losses would be 4.63%, nearly nine times larger than under sectoral prohibitions. The costs are marginally larger than in the data doomsday scenario because they also incorporate additional data management costs related to local storage (which are relatively small).

Overall, the data localisation scenarios show that changes in storage conditions without complementary flow prohibitions have only a small impact on global exports and global GDP. The reasons are twofold. First, data storage conditions generate costs from the greater use of domestic data management services inputs, but these represent only a small share of total firm costs across many sectors. Second, the costs associated with domestic storage conditions are modest according to the OECD-WTO Business Questionnaire.

Figure 4.4. Projected change in global exports and GDP as a result of data localisation measures

Note: The figure displays the projected change in percentage of real exports of goods and services and real GDP at the global level under four scenarios: (1) No data localisation; (2) Horizontal storage condition; (3) Sectoral storage and flow prohibition; (4) Horizontal storage and flow prohibition.

Source: Simulations with the WTO Global Trade Model (version with Melitz market structure as described in Bekkers and Francois (2018[34]).

Impact by level of development

The projected changes in real GDP and real exports under the three data localisation scenarios for the different income groups indicate the following (Figure 4.5).

• The gains under the no data localisation scenario (Scenario A) are much larger for low-income economies than for other economies given the restrictive nature of existing data localisation policies in place and the lack of data flow regulation These could see their GDP increase by over 1% as a result of removing existing data localisation policies.

• Under Horizontal storage conditions (Scenario B) there would be small losses for high-income economies (-0.1% of GDP) for which this would represent a shift to more restrictive conditions, small gains for middle income economies (+0.1% of GDP), and large gains for low-income economies (+1% of GDP) which, for these countries, reflect the impact of moving to less restrictive data localisation than is currently in place.

• Sectoral storage and flow prohibitions (Scenario C) generate the largest losses in high-income economies, because the introduction of these policies would both raise the costs of sourcing intermediate inputs and the costs to export goods and services compared to the current regime.

• The impacts of horizontal restrictions (Scenario D) are the same as reported in Figure 4.2.

Figure 4.5. Projected change in real exports and real GDP by income group from changes in data localisation measures

Notes: This figure displays the projected change in percentage of real exports of goods and services and real GDP under three scenarios: (1) No data localisation; (2) Horizontal storage condition; (3) Sectoral storage and flow prohibition for three income groups: low-income economies (less than USD 1 006), lower middle-income economies (between USD 1 006 and USD 3 955, higher middle-income economies (between USD 3 956 and USD 12 235), and high-income economies (more than USD 12 235) employing the World Bank definition with cutoffs for 2017, the baseline year of the simulation data.

Source: Simulations with the WTO Global Trade Model (version with Melitz market structure as described in Bekkers and Francois (2018[34]).

Impact by region

In terms of the regional projections, only the results for Scenarios A-C are reported (Figure 4.6) since the effects of Scenario D are similar to the data flow Scenario 4 and much larger than the effects of the other scenarios. The positive effects of moving to no regulation (Scenario A) and horizontal storage conditions (Scenario B) is largest in regions with the most restrictive data localisation policies short of ad hoc data flow authorisation policies. These regions include Central Asia (CEA), and Middle East and Northern Africa (MIN), where an easing of data localisation policies leads to reductions in trade costs and increased trust/WTP raising exports and GDP.

Under Scenario C, sectoral storage and flow prohibitions, all regions are projected to lose. Projected changes in GDP are especially large in Europe (EUR), Southeast Asia (SEA) and Oceania (OCE). In these regions, trade costs are projected to increase substantially when the most restrictive data localisation policies are introduced.

Figure 4.6. Projected change in real exports and real GDP by aggregate region from changes in data localisation measures

Notes: This figure displays the projected percentage change of real exports of goods and services and real GDP under three scenarios: (1) No data localisation; (2) Horizontal storage condition; (3) Sectoral storage and flow prohibition for ten aggregate regions: Central Asia (CEA); East Asia (EAS); Europe (EUR); Latin America and the Caribbean (LAC); Middle East and Northern Africa (MIN); North America (NAM); Other Asia (OAS); Oceania (OCE); Southeast Asia (SEA); Sub-Saharan Africa (SSA).

Source: Simulations with the WTO Global Trade Model (version with Melitz market structure as described in Bekkers and Francois (2018[34]).

Notes

← 1.  These are smaller than the costs of geoeconomic fragmentation in the emerging literature (e.g. 5% in Goes and Bekkers (2022[39]) or 7% in IMF (2023[41])). The current scenario only considers trade costs associated with data policies, whereas the other work on geoeconomic fragmentation includes generic increases in both tariffs and non-tariff barriers. Furthermore, the degree of existing fragmentation in approaches to data flows is already high.

← 2.  Using the World Bank’s Classification.

← 3.  The analysis assumes that trade costs within the European Union do not change under the different scenarios because EU Members have the same policies and are not assumed to change their policies vis-à-vis each other.