While Russia is a giant in terms of land and natural resources, it is a medium-sized economy for trade and international investment (Fig. 1); e.g., the 18^th largest trader in the world. The snapshot in Fig. 1 does not reveal the whole story: During the last two decades, the relative importance of Russia in the world economy has fluctuated strongly over time, with fast growth until 2008; turbulence during 2009–2013; and decline from 2014. Figs 2 and 3 show, respectively, Russia’s share of world FDI (Foreign Direct Investment) flows during 1992/1993–2017 (Fig. 2); and Russia’s share of world trade in goods during 1996–2017 (Fig. 3).

Fig. 1.

Russia: Share of the world total, 2016 (%).

Sources: World Development Indicators, UNCTAD for FDI, and trade calculated from WITS/Comtrade data.

Fig. 2.

Russia: Share of world FDI flows (%).

Source: Author’s calculations from UNCTAD data.

Fig. 3.

Russia: Share of world goods trade 1996–2017 (%).

Source: Author’s calculations based on data from WITS/Comtrade.

While FDI flows are more erratic than trade flows, the major pattern over time is the same; with fast growth, turbulence and decline. At the peak, Russia’s share of the world economy was about twice as high as in 2016–2017, for trade as well as FDI.

A main driver underlying the observed pattern was the change in commodity prices. With natural resource rents providing 10.7% of Russia’s GDP in 2017, Russia was the second largest commodity producer in the world,⁵ and fuels represented more than half of Russia’s exports of goods. Because of this, commodity price changes have a strong impact on the Russian economy. Commodity price increases automatically increase the share of commodities in world trade as well as in Russia’s exports. As shown in Melchior (2018, Ch. 2) for 1995–2015, rising commodity prices boosted the value of world commodity trade until 2012, with a fall thereafter. Extending this analysis until 2017, Fig. 4 shows, for the world as a whole, the development of commodity prices during the period, against the share of bilateral two-way manufacturing trade in total world trade.⁶

The close resemblance between the shapes observed in Figs 2 and 3 and the commodity prices in Fig. 4 illustrates that changing commodity prices were a major determinant of Russia’s development. Furthermore, sector shares in trade are strongly influenced by commodity prices: The curve for two-way manufacturing trade in Fig. 4, a proxy for intra-industry trade, is almost a mirror image of the commodity price curve. In the following analysis of Russia, we therefore need to disentangle such nominal effects from true changes in industrial specialization.

Fig. 4.

Two-way manufacturing trade as a share of world trade (%), and commodity prices, 1996–2017.

Note: The commodity price index is for all commodities and energy, based on prices in U.S. dollars. Source: Author’s calculations based on data from WITS/Comtrade, and commodity prices from IMF (https://www.imf.org/external/np/res/commod/index.aspx).

Since nominal export growth for Russia is strongly influenced by commodity prices, it is important to observe that Russia’s imports followed suit (see Fig. 3). Russia has a trade surplus so the share of world exports is higher, but the two curves follow each other in parallel. For imports, the share of manufacturing was higher than for exports, so price fluctuations were likely smaller and more of the variation was due to volume changes. It is likely (although not proven here) that growing export revenues also led to higher imports.

While the average share of Russia in world trade in goods is in the range of 1–2%, Russia’s share of world trade varies considerably across goods and sectors. In some sectors, Russia is a major supplier with a share considerably above the average.⁷

For Russia, the fast trade growth coincided with a shift in the geographical composition of trade, with fast-growing imports from Asia and particularly from China. This is shown in Fig. 5.

Fig. 5.

Russia: Geographical composition of (a) exports and (b) imports, 1996–2017 (%).

Source: Author’s calculations based on data from WITS/Comtrade.

• Western Europe remains the largest trade partner region for Russia, even if its share of trade fell from above 50 to about 40%.
• The share of Eastern Europe in Russia’s trade declined from 18 to 12% for exports, and from 24 to 10% for imports. This is dramatic in the light of the short time period and the integration efforts in the region. The collapse of Russia–Ukraine trade contributed to this development; however, most of the decline happened before the escalation of the Russia–Ukraine conflict in 2014. The relative decline in post-Soviet trade is likely also a transition phenomenon, with a lagged change from historically established trade patterns.
• While Russia’s imports from Asia except China increased nine-fold from 1996 to 2017, imports from China multiplied by 37, ⁸ with China’s share rising from 1.6 to 21.2% of Russian imports. In 2017, Asia including China had a share of 35% of Russia’s imports, slightly below Western Europe. The share of Asia in Russia’s exports also increased — however not like imports — from 14 to 22% for China and the rest of Asia taken together.
• North America accounted for a relatively stable share of Russia’s imports (around 5–7%), but its importance in Russia’s exports declined (from 6 to 3%).

Splitting out Russia from Eastern Europe, and China from Asia/Pacific, Table 1 shows nominal growth from 1996 to 2017 for the matrix of trade flows between the seven world regions. Appendix Table A2 shows the corresponding shares of world trade in 1996 and 2017.

In nominal terms, world trade has grown considerably over time, but the trade of Russia has grown faster, as already evident from Fig. 3. Table 1 shows the change between the end points; Fig. 6 shows the development for world trade (export + imports of goods) and Russia’s trade over the whole period 1996–2017.

Fig. 6.

The nominal growth of world trade and Russia’s trade, 1996–2017 (1996 = 1).

Source: Author’s calculations based on data from WITS/Comtrade.

The nominal value of world trade more than tripled during the period, but the value of Russia’s trade grew even faster, increasing almost six-fold from 1996 to the peak in 2013. At the end of the period, the increase for Russia was about four-fold (equal to the weighted average of the figures for total exports and imports in Table 1).

Table 1 illustrates the spectacular trade growth of China: China’s trade with all trade partners was multiplied by 12 (exports) and 14 (imports). China’s exports to Russia were multiplied by 37, and to the rest of Eastern Europe by 103! Trade in the other directions also grew faster than world trade, but much slower than China’s exports. In spite of this, Russia–China trade has recently been more or less balanced, due to large commodity exports to China from Russia.

Table 1Download as CSV 

Nominal trade growth between Russia, China and major world regions, 1996–2017 (ratios trade₂₀₁₇ / trade₁₉₉₆).

Exporting region	Importing region
	Africa	Asia xChina	China	Eastern Europe xRus	Latin America	Middle East	North America	Russia	Western Europe	World
Africa	5.7	4.8	56.6	13.7	2.9	5.2	1.9	4.5	2.5	3.6
Asia xChina	4.2	2.9	14.6	6.5	3.3	4.9	2.2	6.9	2.6	3.4
China	37.4	9.1		102.6	32.9	32.2	14.3	36.7	14.8	12.5
Eastern Europe xRus	20.7	7.2	14.1	2.7	2.4	7.5	3.9	1.8	9.1	4.8
Latin America	4.9	4.5	34.3	3.8	2.8	6.6	2.6	5.5	2.5	3.6
Middle East	7.0	4.5	44.4	6.7	3.8	7.2	3.6	4.4	4.3	5.2
North America	2.4	1.6	10.0	3.4	2.7	2.9	2.5	3.2	2.2	2.4
Russia	11.7	5.1	8.0	2.8	1.7	7.5	3.4		3.2	3.8
Western Europe	2.8	2.3	12.9	4.3	2.3	3.5	3.1	3.6	2.5	2.6
World	4.5	3.1	14.4	4.0	3.1	5.1	2.9	4.2	2.6	3.1

Note: China and Russia are split out from Asia and Eastern Europe, respectively (indicated by xChina and xRus). Here the left column indicates the exporting region/country, and the top header the importing region/country. The bottom row (far right column) shows nominal growth in the total imports (exports) of each region/country. Source: Author’s calculations based on data from WITS/Comtrade. Ratios based on export and import data differ somewhat, and here an average of the two has been used.

Was the fast growth and reorientation of Russia’s trade merely a switch to new suppliers and trades, or has it changed Russia’s pattern of specialization in trade? In order to check this, we divide trade into manufacturing and other trade. Other trade includes agriculture/food, raw materials, non-ferrous metals and oil/gas.⁹ Fig. 7 shows the share of manufacturing in Russia’s exports and imports, and the share of two-way manufacturing trade in total trade, using the term IIT (intra-industry trade) since it measures the “trade overlap” (see footnote 7).

Fig. 7.

Russia: Share of manufacturing in trade, and share of two-way manufacturing trade in total trade, 1996–2017 (%).

Source: Author’s calculations based on data from WITS/Comtrade.

Given Russia’s large commodity exports, we expect that the share of manufacturing in exports, and the share of two-way manufacturing trade in total trade, have been falling as long as commodity prices were on the rise. This is indeed confirmed in Fig. 7; with V-shaped curves for the two and a turning point around 2011, and a pattern resembling the one observed for global trade in Fig. 4.

A much more dramatic change took place for Russia’s imports, where the share of manufacturing increased significantly, from less than 50 to more than 80%. Hence not only the geographical but also the sectoral composition of Russia’s trade has changed strongly over just two decades only.

These trends suggest a deteriorating trade balance for manufacturing. This can be shown using a simple net export ratio for manufacturing, ranging from minus 1 (only imports) to plus 1 (only exports). While the sector shares in Fig. 7 are affected by commodity prices, especially for exports and two-way trade, the net export ratio for manufacturing largely avoids this commodity price effect. Fig. 8 shows this index.¹⁰

Fig. 8.

Russia’s trade specialization in manufacturing, 1996–2017 (net trade ratios).

Note: Net trade ratios, see explanation in text. Source: Author’s calculations based on data from WITS/Comtrade.

It is evident that Russian trade growth corresponded to a weakening trade balance for manufacturing, with a dramatic change during one decade only. Manufacturing decline could partly be a lagged post-communist transition effect, with old value chains in Eastern Europe being gradually dismantled. The development could also indicate that Russia had a “Dutch disease” syndrome during 2000–2011, although it is beyond our scope here to go in depth on this issue. As described by Corden and Neary (1982), the “spending effect” from higher commodity prices may drive out manufacturing in commodity-producing countries. The reversal of this trend during 2012–2017 could be explained by a similar model, but trade sanctions and import substitution policies might also have played a role. The development observed here is an important backdrop for the “import substitution policies” of Russia from 2014 (Government of the Russian Federation, 2018). While many countries have experienced falling shares for manufacturing due to technological change and weakening trade balances due to the rise of Asia, the change for Russia during 2000–2011 is faster than for other large nations (also ahead of the USA, that had an almost similar development).

This deindustrialization in Russia’s trade played out differently across trade partners, with considerable differences across regions, especially for exports. Appendix Table A3 shows the shares of manufacturing in exports and imports, for the same partner countries and regions.

• Russia’s trade with Western Europe has throughout the period been an exchange of commodities for manufacturing, with no major change.
• For Asia, the pattern has changed dramatically, with a falling share of manufacturing in exports and a rising share in imports. For China, there was a complete reversal of roles, with Russia being the manufacturing exporter in 1996, but the commodity exporter in 2017.
• CIS and Latin America are the only regions where the share of manufacturing in Russia’s exports has been rising over time.

For the simulations, we use the world trade model developed by Melchior (2018).¹¹ Here we provide an overview of the model characteristics and motivation, and the interested reader is referred to the link in footnote 12 for further technical detail. In this article, we present new scenarios and original results, but the model framework is identical to Melchior (2018).

The model is a general equilibrium model of the world economy; with a mathematical solution determining wages, prices and production in all the world’s countries, as well as trade between them. It is a static model without growth, and the numerical simulations are used to find this solution; not to examine trajectories over time. Trade policy is examined by changing trade costs in the model, and comparing the results to the base scenario. Key properties of the model structure are:

• Each country or region is endowed with capital, labor and natural resources, in given quantities.
• Capital K and labor L are fully employed and combined to produce services S (we drop country subscripts for simplicity), that are partly consumed and partly used in the production of tradables X.
• Tradables X are produced combining natural resources or commodities G and services S.
• Commodities G are traded internationally at zero cost, and used as numeraire for measuring prices and costs; with the world price of G equal to one.
• Tradables X is a standard “Dixit–Stiglitz” sector with product differentiation, scale economies and monopolistic competition, traded internationally with real trade costs between countries and regions. ¹² We often call this manufacturing, although in real life tradables could be manufacturing or services.
• Since K and L are not used directly in the production of X, and S is not traded, there is no trade exchange of capital-intensive against labor-intensive goods. But countries with a high K/L ratio are more productive in the production of X, and if they have a trade surplus in X, they must be importing G. The model thereby replicates a key feature of the observed world trade pattern, with large intra-industry trade between rich industrial countries, and exchange of manufacturing for commodities between industrial and commodity-exporting countries (as shown in Melchior, 2018, Ch. 2).

The model solved numerically using MATLAB software. Solving for the wage levels in each country and region, the rest falls into place. The exogenous variables are K, L, G and the matrix of trade costs T, plus various elasticities and cost shares in demand and production functions. The number of manufacturing firms in each country/region is determined endogenously. Countries with large natural resource endowments G may be deindustrialized with zero production of manufacturing X. In the model base scenario, twelve out of 110 countries and regions are deindustrialized with no X production; with seven Russian domestic regions among these. The endogenous and mathematically consistent handling of corner solutions is an original feature of the model. For more technical detail, see Melchior (2018).

While there is a vivid discussion about the existence of a “resource curse” or not (see, e.g., Alexeev and Conrad, 2009; Van der Ploeg, 2011), the model here predicts that natural resources are a blessing in terms of the income they generate; however they may be a curse in terms of deindustrialization, due to the “spending effect” as described by Corden and Neary (1982): Resource income bids up wages and leads to lower production of tradables. Harding and Venables (2013) provide recent evidence to the effect that resource abundance reduces manufacturing exports. As shown in Melchior (2018), commodity-abundant and commodity-scarce countries have most to gain from international trade, since they have too much or too little natural resources, and trade is the solution. Corden and Neary (1982) also include a “resource movement effect” whereby natural resources production affects factor markets and factor prices; this effect is not present in our model since G is an endowment only, ready for use or sale and requiring no further processing. This is clearly a simplification, and in real life, some resource-based sectors such as metal production represent an intermediate category that the model does not capture so well in its current version.

The model is implemented with 110 countries and regions, using data for 2014. Factor endowments are obtained as follows:

• Labor force and population (used for calculating per capita measures) are from World Bank data. We think of this as “raw” labor, and skills should therefore be reflected in the capital measure.
• Natural resource endowments are derived based on World Bank data on natural resource rents as a share of GDP (World Bank, 2006, 2011 and later online data). The inclusion of commodities in the model is important since about half the world’s countries are commodity exporters, and it is important to shed light on whether trade policies affect these countries differently.
• For capital, we draw on the growth literature result that capital-labor ratios are highly correlated with GDP per capita (Caselli, 2005); and there is complementarity between physical and human capital at the country level (for discussions, see, e.g., Parro, 2013; Hsieh and Klenow, 2010). Using recent evidence from the growth accounting literature (Zuleta and Sturgill, 2015), we use the relationship between capital-labor ratios and GDP per capita to derive capital-labor ratios (including human capital) for all countries and domestic regions.

Trade costs have four components:

• tariffs (including preferential tariffs, from the UNCTAD/TRAINS database);
• infrastructure costs (internet users, logistics performance index, costs of business start-up, all from World Bank data);
• export and import costs (costs, documents, time to export and import, from World Bank data);
• distance-related costs including transport costs, derived as a scaling of geographical distance, and assuming that trade costs increase less than proportionally with distance.

The four components enter additively in the overall trade cost mark-up. Appendix Table A4 shows simple averages of total trade costs between Russia and the other world regions, and inside Russia. The scaling of the three latter trade cost components is uncertain, and we use estimates from existing literature, such as Anderson and Wincoop (2004), to make an informed choice. Even if all trade cost components are based on real data, there is uncertainty about the scaling of its components except for tariffs, and this can be better underpinned in the future, based on empirical research. According to the data set used in the simulation model, average trade costs for Russia’s trade with the whole world amount to 55%. Out of these, 5% are tariffs; 7% infrastructure costs; 11% export/import costs; and 31% distance-related or transport costs. Our average trade costs at 55% may be compared to that of Anderson and Wincoop (2004), who found 70%. With more recent data, it is plausible that we have a lower average.

Empirical research using the “gravity model” has confirmed beyond doubt that trade falls with distance (Head and Mayer, 2014). Trade is therefore strongly affected by geography, and we have all reasons to believe that this applies within as well as between countries. In order to capture geography in a better way, a special feature of the model is that large countries are decomposed into regions: Russia, Kazakhstan, China, India, Brazil, USA, and Canada. The motivation is obvious: treating Russia or other large countries as single points, at par with Luxemburg, obviously misses important aspects of geography. There is a growing body of research on regional dimensions of large countries; see, e.g., Autor et al. (2013, 2016) on the USA, or Banerjee et al. (2012) on China. Given the large territory of Russia, we split the country into twelve regions. Appendix Table A5 presents the regional subdivision of Russia, partly using Federal Districts but splitting some of them for geographical reasons. Earlier work, e.g., by Melchior (2010, 2011), see also the survey of Brülhart (2011), demonstrates how domestic regions are affected differently by international trade and integration. As we shall see, this also applies to Russia. Within Russia and other countries that are split into regions, there are still distance- and infrastructure-related trade costs; however, tariffs and export/import costs do not apply inside countries. Hence the model also captures the internal geography of Russia, with considerable trade costs due to distance. As seen from Appendix Table A4, trade costs within Russia are on average 30%; much lower than average trade costs for Russia’s foreign trade.

The model is highly stylized, with three sectors only, and it has not been designed to provide “the exact numbers” about the detailed impact of trade policy, but rather qualitative insight about trade-related mechanisms. The model is simplified, with no government, no financial sector, no currencies etc., so there is no reason to believe that it should match the world perfectly. We therefore do not calibrate the model in order to match the real world exactly, e.g., by adding “wedges” (trade costs, elasticities, etc.) to replicate real world data. The model is “theory with numbers,” and the goal is to capture world trade and geography approximately right. The model replicates income levels rather well, with a correlation coefficient of 96% between observed and predicted income levels in the base scenario. Correlation between predicted and observed trade flows is somewhat lower, at 65% (Melchior, 2018, p. 204). Hence the model is on the right track but further improvement is possible, e.g., by obtaining better estimates for trade costs.

As a reality check for Russia, we may compare observed price levels for Russian regions provided by Gluschenko and Karandashova (2016) with the model predictions in the base scenario.¹³ Figs 9 and 10 show, respectively, correlations between observed price levels and (i) predicted wages (Fig. 9); and (ii) predicted manufacturing prices (Fig. 10).

Fig. 9.

Observed price levels vs. predicted wages for Russian regions.

Source: Observed prices from Gluschenko and Karandashova (2016). Predicted wages from author’s model simulations.

Fig. 10.

Observed price levels vs. predicted manufacturing prices for Russian regions.

Source: Observed prices from Gluschenko and Karandashova (2016). Predicted manufacturing prices from author’s model simulations.

While the good fit between observed and predicted income levels is partly determined by the way the capital-labor ratios are constructed in the data, wages and prices are complex model outcomes so it is interesting to see whether they correspond to reality. In this sense Fig. 10 gives reasons for optimism; the fitted curve has R² = 0.80, so for traded goods, the model seems to be on the right track. For wages, which is a key determinant of price levels in the model, Fig. 9 gives a mixed picture, with the commodity regions Ural, Sakha and Far East (N) being outliers that bring R² down to 0.13. A limitation of the model is that resource income is spent by the factor owners of each domestic region, creating a strong “spending effect.” This is clearly a simplification, given that key firms have headquarters in Moscow and resource income is spent outside the source regions. In future versions of the model, this feature might be modified.

For the interpretation of results, it is useful to observe that for a given trade policy shock in the model, such as trade liberalization, the impact may be reflected in trade specialization effects (some countries exporting more manufactures), and/or wage effects (higher wages in some countries).¹⁴ From international trade textbooks we are used to expect changes in the trade specialization patterns, but as we shall see, wage effects may be equally or sometimes even more important. In the model used here, there are both trade and wage effects, and the result also shed interesting light on price effects of trade policy.

We run the five following scenarios:

• Eastern Europe: Trade integration between Russia and Eastern Europe.
• China: Trade integration between Russia and China.
• EU: Trade integration between Russia and the EU.
• FTA race: Bilateral trade integration between Russia and all other countries.
• Multilateral: Trade integration between all countries in the world.

In this menu, “Eastern Europe” is the most regionalist option; integration with China and the EU go one step towards globalism; and the “FTA race” scenario is full globalism; however, in the form of preferential FTAs, where Russia’s trade partners do not liberalize between them. We therefore also include the fifth “Multilateral” scenario where liberalization is on MFN (Most Favoured Nation) basis and between all countries of the world. While the Multilateral scenario can be promoted via WTO liberalization, Russia could approach the FTA race scenario by means of an ever-expanding set of FTAs.

The four first scenarios are preferential or discriminatory; i.e. trade costs are only reduced between Russia and the partners involved. Given that the tradable/manufacturing sector in the model has scale economies and imperfect competition, and the number of manufacturing firms is endogenous and determined in the model, preferential trade liberalization leads to trade diversion and “production shifting” from third countries to the integrating countries (Baldwin and Venables, 1995). In the Multilateral scenario, trade costs are also reduced between Russia’s trade partners, so the reform is not preferential. This scenario therefore provides a check on the production shifting effect; i.e. how much of the gains from integration are driven by trade discrimination and diversion.

In the scenarios, we make the stylized assumption that all types of trade costs are reduced by 25% between countries involved. We do not ask whether this is feasible or not, or whether some of it has already been undertaken. In order to reduce all these costs in real life, not only tariff cuts are needed, but also infrastructure development, trade facilitation and better transport networks. Cutting 25% of all costs is clearly a significant reform. An interesting issue is whether reductions in distance-related costs have different effects compared to the trade costs that are not spatially dependent (see, e.g, Melchior, 2011). For the sake of brevity, we do not examine this further here; only assume that all types of costs are reduced in the same proportion. For Russia, the distinction may be important due to the vast land area, and this may be addressed in further research.

Given the stylized nature of the model, we are not so interested in the absolute magnitudes but rather the changes induced by the different scenarios.¹⁵ We therefore generally report changes from the base case. Appendix Table A6 reports key results for Russia, China and major world regions for the five scenarios. For Russia, the simulation output is at the level of domestic regions, but we aggregate and present the results first for Russia as a whole, in order to provide an overview across scenarios. Table 2 and Fig. 11 show the average outcome for Russia.¹⁶ Appendix Table A6 shows similar results for all world regions. Later, the same results will be reported for Russian domestic regions.

Fig. 11.

The impact of trade integration for Russia: Change from base case in five scenarios, for key variables.

Source: Author’s results from numerical model simulations.

Table 2Download as CSV 

The impact of trade integration for Russia. Changes from base case in five different scenarios (%).

Variable	Scenario (see main text for explanation)
	Eastern Europe	China	EU	FTA race	Multilateral
Nominal wage	0.22	1.34	0.91	7.31	0.46
Welfare	0.10	0.95	0.77	4.31	2.98
Manufacturing	0.21	1.33	0.89	6.83	0.47
Price level	0.13	0.41	0.16	2.99	–2.44
Nominal GDP	0.19	1.15	0.78	6.27	0.39

Source: Author’s results from numerical model simulation.

In Fig. 11, the four lowest scenarios all represent preferential integration, but with different partners. It is clear that gains from integration with EU and China are of similar magnitude, and several times larger than for integration with Eastern Europe. Even more can be gained from preferential integration with the whole world, as shown by the FTA race scenario.

Hence the results indicate that the potential benefits from regional integration in the post-Soviet space are small compared to the potential gains from integration with more distant world regions. In the real world, this clearly depends on what is feasible: If feasible integration can be much deeper in the neighbourhood, it will change the balance. For tariffs, distant integration is clearly feasible, and with some caution about feasibility, the results suggest that Russia should pursue more FTAs on the global scene.

Given that Russia is a commodity exporter, a potential fear is that trade integration will lead to further industrial decline. The results suggest that the opposite is in fact the case: For Russia, preferential integration leads to considerable growth in the number of firms in the manufacturing sector. Hence the results indicate that commodity traders have no reason to fear free trade. For the whole world, Melchior (2018) also found that trade liberalization tends to promote diversification in commodity-abundant countries. We will revert to this issue when discussing the results for Russian regions, of which some are commodity-rich and others not.

In partial equilibrium, we generally expect that trade liberalization leads to lower prices. In our results for Russia, this is reversed in all the preferential trade liberalization scenarios: here trade liberalization drives up wages, prices and nominal GDP, and the income growth is large enough to generate an overall welfare gain. This is a general equilibrium effect in the model, and illustrates that partial effects may be reversed by the more complex economy-wide interactions. In order to interpret this result, it may be observed from Appendix Table A6 that the wage and price hike in the preferential scenarios applies only to the integrating partners, and is reversed for other world regions. A closer examination reveals that in the preferential scenarios, the total number of manufacturing firms in the world is constant or slightly falling, so trade integration shifts firms and thereby labor demand across countries. Hence production shifting drives the differential price and wage developments in the integrating and non-integrating countries.

Turning to the non-preferential Multilateral scenario, liberalization now also leads to a price level reduction in Russia. Comparing the dark columns for the number of manufacturing firms (see Fig. 11), there is a slight increase for Russia also in this case, but much smaller than in the preferential FTA race scenario. This illustrates the role of “production shifting” due to FTAs. With production shifting gone, manufacturing growth is smaller, and the impact on wages and prices mostly vanished.¹⁷ In spite of this, the welfare gain in the Multilateral scenario is almost as large as in the preferential FTA race scenario, due to the lower prices. Looking at Appendix Table A6, we also observe that the Multilateral scenario is more equitable: all world regions gain in welfare. There is an increase in the global number of manufacturing firms, and there is some manufacturing growth in most world regions. This growth is strongest in some developing regions, notably Africa and Latin America. Eastern Europe is here in third place suggesting that they might also have an interest in globalism.

Russia is an interesting country in this context for many reasons; one is its geography; another is that some domestic regions are pure commodity exporters while others are diversified. As shown by Fig. 12, large parts of Russia are resource-based, with zero predicted output of manufactures.

Fig. 12.

The share of tradables in GDP for Russian regions, according to base scenario.

Source: Author’s results from numerical model simulations.

In the base scenario of the model, the predicted manufacturing production is zero for seven of the twelve Russian regions. The largest domestic regions for manufacturing/tradables are the Central Federal District (with Moscow), North West (1) (with St. Petersburg), followed by the Southern Federal District, West Siberia and Volga. Some of the commodity regions have very high GDP per capita (Ural, Sakha); and the South has a much lower income level than other manufacturing regions.¹⁸

If a domestic region becomes deindustrialized, it will export all its commodity endowment (since nothing is used for domestic tradables production), and import all its consumption of tradables. The region’s capital and labor endowment will be used entirely in the services sector, and by assumption the commodity income will be redistributed for consumption in the given domestic region. Foreign trade effects are then just a matter of prices for commodities versus tradables, since the markets for capital and labor are unaffected by trade. For the commodity regions, the terms of trade are therefore a key issue. With high commodity prices, the country or domestic region becomes rich and can import more for its consumption.

The cut-off point for becoming deindustrialized is determined endogenously in the model, so the number of deindustrialized countries and domestic regions can vary across scenarios. In the base scenario, twelve countries or domestic regions are predicted to have zero tradables production. In addition to our seven Russian regions, the group includes West Kazakhstan, Azerbaijan, Arabia (a group of seven Middle East countries), Iran and the Chinese Province of Mongolia. In our simulations, the number of deindustrialized countries is still twelve in the Eastern Europe scenario, but reduced to eleven in the China, EU, FTA race and Multilateral scenarios. The domestic region becoming diversified due to trade integration is East Siberia, which establishes some tradables production (although small). This is another illustration of the diversifying impact of trade liberalization for commodity-exporting countries suggested by the model.

Turning to the detailed simulation results, however, it is not mainly the diversification effect that raises Russia’s overall manufacturing production, but production growth in the key manufacturing regions. For brevity, we present only two scenarios; the preferential FTA race scenario and the Multilateral scenario. Compared to the FTA race scenario, the other preferential scenarios (Eastern Europe, China, EU) are similar but quantitatively smaller (except that East Siberia remains deindustrialized in the Eastern Europe scenario). In Appendix Table A7, we show domestic regional results for all the five scenarios. In order to facilitate reading and visualisation, we have sorted domestic regions with the commodity regions at the bottom of the table as well as the diagrams.

Results for the preferential FTA race scenario is shown in Fig. 13. Here we show wages, welfare, manufacturing and price level changes.¹⁹ The results demonstrate again that preferential trade liberalization promotes manufacturing, and for one region — East Siberia — this influence is sufficient to cross the borderline and become diversified.

For the other six commodity regions at the bottom of Fig. 13, manufacturing production remains at zero, and the nominal wage is also unchanged since it is not affected by trade. However, the price level reduction due to cheaper imports leads to a welfare gain — the two are of similar magnitude and with opposite signs.

Fig. 13.

FTA race scenario: Predicted impact for Russia regions.

Source: Author’s results from numerical model simulations.

For the manufacturing regions, there is not a price level reduction but — as seen for Russia as a whole — rather the opposite: wages increase significantly and therefore also the price levels. In spite of this, the income gain is large enough to generate a welfare gain that is larger than for the commodity regions.

Considering Russia as a whole, the bulk of manufacturing growth takes place in the domestic regions that were already diversified in the base scenario. Here there is an interesting economic geography pattern, with higher increases in manufacturing for Volga and West Siberia. The exact reasons are not so easy to pin down, since this is the result of complex interactions in the model. One possible driver is higher demand from neighbour countries and domestic regions, especially the Russian commodity regions. Another explanation can be geography: Volga and West Siberia are centrally located regions and it may be the case that external liberalization promotes development in this area. In spite of the higher manufacturing growth in these two domestic regions, the welfare gain is even higher for the key manufacturing regions: Central and North West (1). The reason is the higher weight of the tradables sector in their economies.

In Appendix Table A7, results for the three other preferential scenarios (Eastern Europe, China, EU) are shown. The results are similar to the FTA race scenario, just quantitatively smaller, and with a similar ranking as seen for Russia as a whole. This is surprising since one might expect, e.g., that integration with China should be better for the east of Russia; and integration with the EU should be better for the west. But the ranking in terms of manufacturing growth across domestic regions is similar across the preferential scenarios, with Volga and West Siberia as the winners for manufacturing across all scenarios. One interpretation is that since the geographical pattern is stable across scenarios, it may be demand from neighbour regions that drives this positive outcome in the central parts of Russia. However, more research is needed to obtain a firm conclusion on this.

While it is beyond the scope of this article to undertake a further empirical analysis of the Russian regions, the FTA race scenario illustrates some key phenomena that should be taken into account in empirical research on the impact of FTAs:

• Industrial and commodity regions differ fundamentally with respect to the production and price effects of integration.
• Economic geography creates spatial differences with respect to the impact of integration, so the effects differ across regions due to their location.

These two implications apply similarly to the analysis of preferential trade integration among countries.

Turning to the non-preferential Multilateral scenario, we have seen from the country-level results (see Fig. 11) that it is qualitatively different from the preferential ones. Fig. 14 shows the underlying results at the regional level within Russia, in stark contrast to the preferential scenario in Fig. 13.

The difference is particularly marked for the industrial regions, since multilateral liberalization wipes out the production-shifting to Russia seen in the preferential scenarios, and the corresponding wage- and price-driving impact of manufacturing growth. Now the price and welfare effects are similar for all Russian regions, with a welfare gain that is mainly due to cheaper (or more diversified) imports, and the resulting lower price level. Hence Multilateral integration is more equitable across regions, but preferential FTA race integration is Paretoimproving since the commodity regions get their price level reductions also in this case.

As noted earlier, the Multilateral scenario leads to global growth in the number of manufacturing firms (due to the larger overall reduction in trade costs). Fig. 14 shows that Russia also gets a share of this, with manufacturing growth in the diversified domestic regions, and diversification of East Siberia also in this case. Again, there is more industrial growth in the Volga and West Siberia regions.

Fig. 14.

Multilateral scenario: Predicted impact for Russian regions.

Source: Author’s results from numerical model simulations.

Hence in the Multilateral scenario, the industrial and commodity domestic regions are not as different as they were in the preferential setting, but the threshold between deindustrialization and diversification is still at work, and the manufacturing effects still apply to the diversified regions. The results of this section provide a conceptual framework and hypotheses for further empirical work, and the results concerning price and wage effects suggest that these variables are of great interest in empirical work on trade integration, be it at the country level or for domestic regions. As shown earlier, there are considerable price level differences across Russian regions, and the predicted price levels for tradables correspond well to the real observations. Hence it is hoped that the model simulations presented here provide a useful framework for further empirical analysis.