Global value chains: Efficiency and risks in the context of COVID-19

 The degree of concentration can amplify or dampen international shocks

While competition based on comparative advantages (e.g. resource endowments and ownership of technology or know-how) determines countries’ specialisation and shapes the geographical structure of GVCs, the subsequent organisation of production in GVCs often ends up being very concentrated. A small number of firms that both import and export dominate GVCs.6 This concentration, if associated with anti-competitive behaviours, can potentially undermine the initial efficiency gains. More generally, concentration typically amplifies volatility.

Supply chains characterised by low diversity of suppliers or buyers can indeed increase the probability of disruption and can magnify the propagation of shocks. For example, for some parts of line telephone equipment, three‑quarters of world supply is provided by three countries: China, Viet Nam and Korea. Downstream firms relying on these specific inputs have not much leeway to switch towards new supply sources in case of shortage, at least in the short term. Concentration of suppliers reduces the scope for diversifying away from a supplier facing disruptions. Likewise, on the demand side, reliance on too few customers may be a source of fragility for suppliers. In this case, a demand shock in a destination market cannot be buffered and may have large consequences on upstream firms serving this market. However, concentration is often also a manifestation of comparative advantage, specialisation and economies of scale, and it is not clear that it should be reversed, or what would be the wider implications of the policy tools and incentives used to do so.

In many markets, exports tend to be concentrated in a few supplying countries (Figure 1, Panel A). Import destinations, in contrast, tend to be more diversified. For a commonly used measure of concentration – the cumulative share of the top-5 biggest countries (CR5) in the world market – the distribution of export concentration across products is tilted to the right compared to the concentration of imports. This means that the supply side is quite concentrated, with a few countries specialising in producing and selling abroad, but the supply ultimately serves final demand in a diversified range of countries.

Focusing on product groups shows that computer and phones exhibit by far the highest degree of concentration in the global export market (Figure 1, Panel B). On average, 79% of the market share of computer‑related products is accounted for by five countries only and the top‑five countries account for 75% of world exports of phones. In general, the difference in concentration of exports compared to imports is smaller for intermediate goods, highlighting that an important characteristic of GVCs is that relatively few GVC hubs rely on multiple source countries. This is also supported by the finding that supply of intermediate goods is less concentrated than supply of final goods, pointing out that possible bottlenecks may occur more downstream of the value chain, while a broader range of intermediate suppliers can be drawn upon in case of a disruption.

 

Figure 1. Exports are concentrated on fewer countries than imports

Note: Using bilateral exports and imports flows from the BACI database, indicators of concentration have been constructed for 2018, at the 6‑digit level of the Harmonised system (HS) classification. For each of the 4726 products traded internationally, an exporting country is defined as a supplier and an importing country is a customer. The concentration ratio of the top five supplier-countries (CR5) gives the cumulative market share accounted for by the 5 biggest countries to total world exports (for a given HS6-product). In Panel A, the distribution reports this statistics for 4726 HS6 products. “Fraction of products” represent the percentage of traded products with a given concentration measure. By construction, the sum of these fractions over all bins is equal to 100%. In Panel B, boxplots represent the value of the weighted first quartile, median and third quartile of CR5 across products belonging to each end-use category. The diamonds indicate the weighted mean.

Source: BACI 2020 version and OECD calculations. For details on the database construction, see Gaulier and Zignago (2010[6]), “BACI: International Trade Database at the Product-Level. The 1994-2007 Version”.

 Countries are exposed differently to foreign supply and demand shocks in GVCs

While the concentration measures computed on the basis of detailed trade data can point to exposure to shocks and bottlenecks at a fairly granular level, they are based on gross trade figures and do not account for the key GVC characteristic: only a part of value is added by the exporter — the rest comes from other GVC participants which are often located abroad. Typically, a relatively higher share of foreign value-added from foreign input providers (so‑called “backward” GVC linkages) can indicate a higher exposure to foreign supply shocks affecting vendors of raw materials and intermediates. Conversely, a higher reliance of exports of a given country on demand from foreign countries (so called “forward” GVC linkages) can mean higher exposure to demand shocks coming from final consumers or distributive services abroad.

Connectedness to global markets enables smaller economies to reach a bigger customer base, but they are also more exposed to demand shocks in GVCs than large economies. This is particularly the case for manufacturers in Central and East European countries that provide inputs into the European automotive industry, or for Irish firms supplying the global information technology sector. These economies export between 60% and 80% of the total amount of value added produced domestically (Figure 2, Panel A).

The exposure to foreign supply shocks also tends to be larger for smaller countries where production depends on foreign inputs to a larger extent (Figure 2, Panel B). The European Union, the United States, in China, and other Asian countries provide high portions of foreign inputs used in manufacturing across the world. Still, key suppliers of foreign inputs to manufacturing (potential sources of shocks) are distributed relatively more evenly than suppliers to the business services sector where intermediate services tend to be sourced mainly from Europe and the United States, thus reliance on Asia is lower.

 

Figure 2. Small economies are most exposed to shocks to the manufacturing sector

Note: the degree of exposure to a demand shock is computed as a share of domestic value added in foreign final demand in total domestic value added in 2015. It includes exports of final goods or services directly reaching foreign markets, but also intermediates reaching foreign final consumers only indirectly. The degree of exposure to a supply shock is computed as a share of foreign value added in gross output of the sector.Other Asia includes Japan, Korea, India and ASEAN countries.

Source: OECD TiVA database and OECD calculations.

 Some countries and industries are very central in the GVC network

Moving from a country to a global perspective, some countries and industries might have more systemic implications because of their position in the global production network. Measures of centrality (Criscuolo and Timmis, 2018[7]) capture this influence in the GVC network. Countries or industries are central when they are highly connected with other major hubs, conversely they are peripheral when they reveal weaker trade linkages. Thanks to their pivotal position across global production networks, central hubs become particularly influential suppliers or customers in GVCs and therefore they can play a key role in both upstream and downstream transmission of shocks.

Since the mid-2000s, the centrality of China as the main manufacturing hub in several sectors has grown significantly, both as a source and as a destination of value added.7 In some industries China displays the highest degree of centrality for the year 2016 (the latest year with data) and the largest increase in the centrality metrics since mid-2000s (Figure 3, Panel A). Computers and electronics manufacturing has also undergone profound changes, with the core of the network shifting from the United States and Korea towards China, which currently is both the most central buyer and supplier of inputs in the whole industry. The German and US motor vehicle industries remain two of the most central manufacturing hubs globally.

In the service sector, the centrality metrics for 2016 confirm the central position of high-income economies, such as the United States, the United Kingdom, Germany and France as key hubs (Figure 3, Panel B). The Unites States, in particular, is the most central provider of business services, including financial and insurance services, legal and accounting activities, wholesale and retail trade, and R&D. The most sizeable changes in the structure of networks are represented by the rise of the French R&D and business support industry and the decline of the UK financial and insurance sector.

 

Figure 3. Top ten most centrals hubs

Note: Total centrality is computed as average of forward and backward centrality. Forward centrality metrics capture the importance of a country or a sector as a seller of value added in intermediates for the production of exports of a specific partner while backward metrics measure the importance of a country as a buyer of value added in intermediates for the production of its own exports. The manufacturing sector excludes construction while the service sector excludes electricity, gas, and water supply services. The “Rest of the world” has been excluded from the chart as it represents an aggregate of several heterogeneous countries whose distinct trade flows are not disentangled.

Source: OECD (2018) Inter-Country Input-Output (ICIO) database (http://www.oecd.org/sti/ind/inter-country-input-output-tables.htm) and OECD calculations.

 Comparing “interconnected” versus “localised” regimes

To do that, two stylised versions of the global economy (regimes) are explored. The interconnected regime represents production fragmentation in GVCs, much as is seen today, accounting also for the effects identified to-date of the demand and supply changes during the COVID-19 crisis. The localised regime reflects a situation where, on top of the effects of the COVID-19 crisis, incentives to source inputs from abroad are reduced through two mechanisms. First, there is a global rise in import tariffs on all traded products to 25% in all regions.9 Second, this is combined with national value-added subsidies equivalent to 1% of GDP directed to labour and capital in domestic non-services sectors in each country in order to mimic rescue subsidies that favour local production.10 In addition, it is also assumed in the localised regime that firms are more constrained in switching between different sources of inputs, making international supply chains more “rigid”.11

The simulated policy assumptions in the localised regime create strong incentives to increase domestic production and rely less on international trade. National value-added subsidies combined with import tariffs increase prices of imported products relative to domestic ones and make producers in all countries and in all sectors less reliant on foreign inputs for their production and exports than in the interconnected regime. The two regimes can then be compared along two dimensions: their efficiency (mean levels of economic activity) and their relative stability (deviations in levels of economic activity from the mean in response to shocks).

 Localisation of GVCs would add further GDP losses to the economic slowdown caused by the pandemic

First, the analysis explores effects of shifting towards the localised regime on key economic variables such as trade, production and real GDP, i.e. the overall cost of making the supply chains more ‘local’ or the ‘efficiency’ side of the purported efficiency/stability trade-off. The results show that the localised regime, where economies are less interconnected via GVCs, has significantly lower levels of economic activity and lower incomes in all economies. This suggests that greater localisation of value chains would add further GDP losses to the economic slowdown caused by the pandemic.

The shift to the localised regime is estimated to decrease global real GDP by more than 5% relative to the interconnected regime (Table 1). Reductions in economic activity are significant across all regions and countries, and in some of them the percentage cuts reach double digits. This underscores the gains in output and jobs around the world from international specialisation that have been realised over the past decades, including through reductions in barriers to trade. Import and export demand fall in real terms proportionally more than real GDP and domestic production, with the result that, in the localised regime, exports and imports come to account for smaller shares of GDP.

 Would localisation of GVCs result in more stability in the face of supply chain shocks?

To explore how the interconnected and the localised regimes compare in terms of the propagation of, or insulation from, shocks, a stylised set of ‘supply chain’ shocks is explored, based on a 10% increase in costs of bilateral exports and imports between a given region and all other countries. Since countries experiencing the shocks are both sources and destinations of intermediate and final products, the set of shocks mimics the kind of disruptions experienced during lockdowns to contain the COVID-19 pandemic, when transport, labour and logistics disruptions affected both exports and imports of different products to a similar extent.

The shocks are country-specific, as was the case with COVID-19, where supplies across many industries were initially disrupted in China and subsequently in other countries. The shocks are sector‑generic – that is, they are applied equally across all sectors to capture the fact that sectors source from a range of different industries (e.g. car producers do not just source from other firms in the car industry, but also from other sectors). The chosen shock characterisation is stylised, and other more specific types of shocks could be usefully studied in the future, but it features the main characteristics of supply-chain disruptions during the COVID-19 pandemic.

In this context, the stability of an economic variable, such as real GDP, is calculated for each country as the absolute value of an average percentage deviation from the initial base level of this variable (i.e. from the base level in either the interconnected, or the localised, regime) across all country-specific shocks, including the one originating in the domestic economy.

 Localised supply chains are found to be more ‒ not less ‒ vulnerable to shocks for most countries

The results of the simulations show that contrary to some of the claims in the general debate on risks in GVCs, in the localised regime, shocks do not result in a significant increase in the stability of GDP, production and consumption relative to the interconnected regime. On average, the stability of real GDP, real production and real consumption declines (Table 2). This is because the localised regime offers generally less flexibility for adjustment in the face of shocks. Domestic markets need to shoulder most of the adjustment pressures and domestic prices and quantities have to adjust relatively more, which leads to less stable consumption for most countries in the localised regime, more so than real GDP and production. Put differently, trade helps smooth shocks to supply of globally consumed products. In the interconnected regime, domestic shocks account for about half of the variation in real GDP due to shocks, illustrating that domestic shocks may be at least as important as a source of instability as foreign shocks. However, in the interconnected economies regime, part of the adjustment is carried by international markets and benefits from diversification of supplies (i.e. the ability to switch towards suppliers of intermediate products which have not been affected by a spike in trade costs).

However, these averages mask a certain heterogeneity across specific countries (Figure 4). For 16 out of 22 countries or regions included in the model, real GDP stability falls when all shocks are considered. Countries affected the most negatively tend to be located upstream in GVCs. Countries such as the Russian Federation, Argentina, Brazil and other Latin American countries lose more than one percentage point in terms of real GDP stability in the localised regime when all shocks are considered. They tend to provide inputs into production for other countries and the concentration of their economies and trade patterns around natural resources and agriculture results in less adjustment domestically because of a less diversified economy that relies on less mobile factors of production.

 

Figure 4. In the localised regime, shocks also result in lower levels and lower stability of real GDP for most countries

Note: All changes in variables are relative to the level of the interconnected regime base scenario which is set to equal 100. Blue dots show the base in the given regime relative to the interconnected base, and whiskers show average deviations for negative and positive trade cost shocks.

Source: OECD METRO database and simulations.

 Selected downstream GVC participating countries could gain marginally in terms of stability in the localised regime…

While the stylised design of the exercise makes the quantitative estimates only exploratory, in the six economies that gain marginally in terms of real GDP stability in the localised regime, this comes at the cost of a much lower level of economic activity. This group includes France, Korea, Mexico, Turkey, the United States, and the region of Southeast Asia. For example, stability is estimated to improve the most in Korea as this country ‘saves’ three‑quarters of a percentage point of variation in real GDP in the localised regime. Still, Korea would experience significant reduction in the level of real GDP (by 7.4%) from shifting to the localised regime. The other five countries typically gain about one-fifth of a percentage point in terms of real GDP stability, which can also be seen as small compared to the costs of switching to the localised regime, which in all cases exceed 5% of real GDP.

The situation of these economies in terms of stability gains in the localised regime can be explained partially by the fact that these are some of the most open and GVC-integrated economies in the interconnected regime, and that they tend to be located downstream in GVCs. On the one hand, they are still more open in the localised regime than many other economies, meaning that adjustments in international markets still help them better adjust to shocks. On the other hand, these economies also experience the largest reductions in trade (and thus sourcing of intermediates from abroad) as well as in incomes when shifting to the localised regime (Figure 5). Thus, the countries that could gain on stability in the localised regime pay the highest cost in terms of the level of real GDP when switching to the localised regime.

 

Figure 5. Those countries that could have lower GDP variation in the localised regime would pay the highest cost in terms of the level of real GDP when switching to the localised regime

Efficiency and stability effects of shifting to the localised regime

Note: for details of methodology used see Arriola et al. (2020[5]).

Source: OECD METRO database and simulations.

 But this would come at a high efficiency cost

In a way, these results indicate the existence of a trade-off between efficiency and exposure to shocks in GVCs for a limited number of countries. But already in this stylised analysis this trade-off does not suggest straightforward incentives for re-localisation. First, the trade-off is relatively steep: several percentage points of GDP would have to be sacrificed in a shift to a localised regime in order to increase its stability by a fraction of a percentage point, in the face of relatively significant trade cost shocks. Second, countries relying the most on GVCs for foreign inputs into their production, while gaining potentially the most in terms insulation from shocks in the localised regime, would have to sacrifice higher portions of incomes to gain such security. Third, and related to the above points, countries located upstream in GVCs would lose out relatively more from switching to the localised regime both in terms of level of income and of its stability.