Abstract This article provides an empirical analysis of the disruption of (neo)classical/liberal globalisation during the Great Depression and the Great Recession, identifying challenges to existing knowledges and approaches, in particular the mainstream analysis that studied the Great Recession in isolation, treating it as a special case and ignoring political covariates of deglobalisation processes. The econometric analysis (an unbalanced panel of 31 countries in the 1928–1932 and 128 countries in 2008–2012) does not find an impact from the level of development on deglobalisation. The manufacturing import share is significantly associated with stronger deglobalisation (this effect is stronger in the 1930s than in the 2000s). The political system is highly significant in both 1930s and 2000s, but the sign of its impact is opposite. In the 1930s, autocratic rule and dictatorship are associated with stronger deglobalisation; in the 2000s, democracy is associated with stronger deglobalisation. Introduction The world trade collapse that started in October 2008 and reduced global trade by about 20% in only a few months’ time remains one of the most puzzling and still widely debated economic phenomena of the last decade.1 After the recovery from its global collapse, world trade slowed down and triggered a process by which the share of international economic activities in the economy has been reduced (which I see as an indicator for deglobalisation). To current observers, the 2010s in this respect may appear to be unique, especially in view of the Trumpism and Brexitism that emerged in 2016/2017, and some have suggested that especially democracies could be vulnerable to such anti-globalist tendencies. According to Murshed (2017), rising inequalities and a sense of relative deprivation perceived to be due to globalisation have intensified conflicts between ruling elites and dissident groups, and these tensions resulted in both a majority in the referendum vote for Brexit (an unprecedented withdrawal from a supranational regional integration framework) and the election of a US president with an explicit “America first” agenda. These events are unique for our generation indeed, since the world in which we grew up was a world in which globalisation was the determining economic trend; from a wider historical perspective, however, the 2000s–2010s fit the stylised facts of the trade and investment developments in the 1930s quite well. Important underlying tendencies in the 1930s are the erosion of the hegemon’s position (Britain), the collapse of the dominant European currency arrangement (Gold Standard) and a cluster of trade enhancing factors, including key technological innovations, long run reductions of trade costs, new modes of transportation and communication and the emergence of previously peripheral countries in the global trade system (‘recent capitalist countries’) (van Bergeijk, 2010, 10–12). A similar pattern can be discerned surrounding the Great Recession, where the hegemon USA is challenged by China, EMU has been under pressure, 3D printing and internet connectivity are fundamentally changing what goods and services are tradeable and traded (and how) and, finally, emerging markets including the BRIICS countries move to the centre of the global trade system. One important methodological point that this article makes is that the phenomenon of the occurrence of deglobalisation as a consequence of a collapse of world trade is more general than assumed by those who only study relatively recent developments, and that it is therefore pertinent to study the two major trade collapses and their aftermath simultaneously.2 I have shown elsewhere (van Bergeijk, 2017) that the mechanism underlying the development of trade flows during the initial phase of the world trade collapses in the 1930s and the 2000s is comparable, but also that statistically significant differences emerge in the mid-term. While the duration of the 2008/2009 world trade downturn on average was shorter and the recovery to pre-collapse levels was much quicker than during the previous world trade collapse that was an important element of the Great Depression of the 1930s, a comparable process of deglobalisation was set in motion by the two world trade collapses. This article builds on this earlier work but more accurately engages with deglobalisation, as it does not deal with an indirect measure (that is, the development of trade flows), but instead directly focuses on the development of the share of trade in GDP.3 One commentator4 has rightfully pointed out that the discussion on (de)globalisation is arguably about structural change and possibly not so much about changes to trade that refer to once-off shocks like the Great Recession or Great Depression. This raises the issue whether these events can indeed be used as indicators of deglobalisation or should be seen as recurrent cycle-related phenomena (either the business cycle; see Ikeda, 2018, or accompanying the long wave; see van Bergeijk, 2013a, 140–144). The structural break in globalisation in the 1930s is well documented (see, for example, van Bergeijk and Mensink, 1997 and van Bergeijk, 2009a), but for a final verdict on the Great Recession and World Trade Collapse of the 2000s probably a longer time frame is necessary. With this caveat in mind, Figure 1 at least suggests that the world trade collapse and world trade slowdown have led to a structural break in the upward trend in the ratio of trade to GPP (Gross Planet Product) that the International Monetary Fund (IMF) for the foreseeable future (up till and including 2022) forecasts to remain below its previous peak level in 2008. Figure 1. View largeDownload slide Openness of the world economy (trade volume goods and services to GPP* ratio in percentage, 1960–2022). GPP is Gross Planet Product (see van Bergeijk, 2013a). Sources: World Bank, World Development Indicators and IMF World Economic Outlook October 2017 Database, accessed 16 November 2017. *GPP is Gross Planet Product (see van Bergeijk, 2013a). Figure 1. View largeDownload slide Openness of the world economy (trade volume goods and services to GPP* ratio in percentage, 1960–2022). GPP is Gross Planet Product (see van Bergeijk, 2013a). Sources: World Bank, World Development Indicators and IMF World Economic Outlook October 2017 Database, accessed 16 November 2017. *GPP is Gross Planet Product (see van Bergeijk, 2013a). In this article, I will use the change in the ratio of imports to gross product to measure (de)globalisation. This is a crude approximation, because (de)globalisation also refers to capital flows (including bank lending, portfolio investment, mergers and acquisitions, Foreign Direct Investment, remittances and development cooperation) and, moreover, to social and political forms of integration (including tourism, migration, cultural and personal exchanges, membership of international institutions, Treaties, peace-keeping missions etc.; see Dreher, 2006). The choice for crudeness rather than sophistication is simply a consequence of limited data availability. According to my definition, a decreasing import share in gross product indicates deglobalisation. The recent Great Trade Collapse (Baldwin, 2009) and the ensuing Global Trade Slowdown (Hoekman, 2015; Timmer et al., 2016) have stimulated a lot of research.5 Unfortunately, the mainstream analysis studies these phenomena without much consideration of its historical precedents. Whereas research on the world trade collapse in the 1930s in the past by necessity studied that case as a single, unique event, most researchers of the world trade collapse of 2008–2009 by choice dealt with periods starting in the 1950s—also when the study pretended to offer a ‘historical perspective’ (Freund, 2009 is an example).6Table 1 provides an overview of the empirical literature that deals with world trade collapses (and their aftermath; that is, the associated phase of deglobalisation). These studies are typically based on empirical analyses of post-Second World War data only (less than 15% of the studies deal with the Great Depression and the Great Recession simultaneously, and these studies are unfortunately descriptive, based on very preliminary data or deal with only a handful of bilateral trade relations in the 1930s, so that it is difficult to see how this can be generalised).7 The literature thus mainly compares the usual fluctuations in international trade to an exceptional trade collapse and derives stylised facts for one of the two historical events. This amounts to treating a world trade collapse and its aftermath as a case study, and this clearly has methodological and analytical costs: simultaneously examining natural experiments such as provided by the two trade collapses and their aftermath may provide useful lessons for future world trade collapse events. Table 1. Empirical studies on world trade collapse(s) and aftermath. Author (year) Period(s) and geographical level of observation Great Depression (1930s) Great Recession and Global Trade Slow Down (2000s–2010s) Altomonte et al. (2012) NO France Aslam et al. (2017) NO 169 countries Bénassy-Quéré et al. (2009) NO 18 world regions Behrens et al. (2013) NO Belgium van den Berg and Jaarsma (2017) NO The Netherlands van Bergeijk (2013b) NO 45 countries van Bergeijk (2017) 34 countries 173 Brakman and van Marrewijk (2017) NO 72 countries Campbell et al. (2009) 27 countries 27 countries Cheung and Guichard (2009) NO World Chor and Manova (2012) NO USA (bilateral trade) Eaton et al. (2011) USA (8 trade partners) 22 countries Eichengreen and O’Rourke (2009)a World Trade World Trade Estevadeordal et al. (2003) 28 countries NO Freund (2009) NO World Haddad et al. (2010) NO Brazil, EU member states, Indonesia, USA (imports) Haugh et al. (2016) NO World Hong et al. (2010) NO 21 Asian developing countries Jacks et al. (2011) 130 country pairs NO Levchenko et al. (2010) NO USA (bilateral trade) Madsen (2001) 17 major trading nations NO van Marrewijk (2009) NO 29 Countries OECD (2009) NO World Paravisini et al. (2015) NO Peru (bilateral trade) Robertson (2009) NO Mexico Tamminen (2017) NO Finland Tanaka (2009) NO USA, Japan Wagner and Gelübcke (2014) NO Germany Author (year) Period(s) and geographical level of observation Great Depression (1930s) Great Recession and Global Trade Slow Down (2000s–2010s) Altomonte et al. (2012) NO France Aslam et al. (2017) NO 169 countries Bénassy-Quéré et al. (2009) NO 18 world regions Behrens et al. (2013) NO Belgium van den Berg and Jaarsma (2017) NO The Netherlands van Bergeijk (2013b) NO 45 countries van Bergeijk (2017) 34 countries 173 Brakman and van Marrewijk (2017) NO 72 countries Campbell et al. (2009) 27 countries 27 countries Cheung and Guichard (2009) NO World Chor and Manova (2012) NO USA (bilateral trade) Eaton et al. (2011) USA (8 trade partners) 22 countries Eichengreen and O’Rourke (2009)a World Trade World Trade Estevadeordal et al. (2003) 28 countries NO Freund (2009) NO World Haddad et al. (2010) NO Brazil, EU member states, Indonesia, USA (imports) Haugh et al. (2016) NO World Hong et al. (2010) NO 21 Asian developing countries Jacks et al. (2011) 130 country pairs NO Levchenko et al. (2010) NO USA (bilateral trade) Madsen (2001) 17 major trading nations NO van Marrewijk (2009) NO 29 Countries OECD (2009) NO World Paravisini et al. (2015) NO Peru (bilateral trade) Robertson (2009) NO Mexico Tamminen (2017) NO Finland Tanaka (2009) NO USA, Japan Wagner and Gelübcke (2014) NO Germany aRegularly updated column at vox.eu. View Large The goal of this article is not to compare two identical twins, but to compare two events that relate to the same phenomenon and have differences and similarities. It is, indeed, very important that the events themselves are not identical because that allows us to learn more about deglobalisation in the wake of world trade collapses. Figure 2 compares the two events for a set of 22 countries where all data on dependent and explanatory variables are available for both periods. While the correlation coefficient is low (R2 = 0.12), so that country experiences in the 1930s and the 2000s can be quite different, Figure 2 illustrates that deglobalisation is a significant and widespread phenomenon in the 1930s and 2000s. An important difference between the 1930s and the 2000s is the comparative difference in the strength of the deglobalisation process: in the first 3 years of the Great Depression of the 1930s, the import to GDP ratio decreases in this sample of 22 countries on average by 17%, while this decline is 31% on average in the 2000s. Figure 2. View largeDownload slide Index numbers import to GDP ratio at constant prices for 22 countries (1928–1932 versus 2008–2012)*. Country coverage: Argentina, Australia, Austria, Brazil, Canada, Chile, Denmark, Finland, France, Germany, India, Indonesia, Italy, Japan, Mexico, Netherlands, New Zealand, Norway, South Africa, Sweden, Switzerland, UK and USA. *Japan is not included although all data are available because it does not deglobalise according to our measure: Japan’s GDP contracts stronger than Japanese imports. Figure 2. View largeDownload slide Index numbers import to GDP ratio at constant prices for 22 countries (1928–1932 versus 2008–2012)*. Country coverage: Argentina, Australia, Austria, Brazil, Canada, Chile, Denmark, Finland, France, Germany, India, Indonesia, Italy, Japan, Mexico, Netherlands, New Zealand, Norway, South Africa, Sweden, Switzerland, UK and USA. *Japan is not included although all data are available because it does not deglobalise according to our measure: Japan’s GDP contracts stronger than Japanese imports. By focusing on either the Great Depression or the Great Recession, the mainstream literature by implication treats the phenomenon of world trade collapse as a unique event and actually ‘such events can only be explained historically as they defy the laws of economics’ (Rothermund, 1996: 1). To address this methodological weakness, this article studies the correlates of deglobalisation in the slipstream of the world’s two major trade collapses. In particular, it fills a gap in the mainstream literature by providing the first comprehensive review of these developments from a comparative perspective on the pace and direction of deglobalisation during the Great Depression and the Great Recession. My succinct econometric model (a quasi-postulated reduced form equation) explores the substantial amount of cross-country variations in deglobalisation by means of an unbalanced panel (31 economies and up to 128 economies during the Great Depression and the Great Recession, respectively). Due to the fact that the empirical investigation covers two very distinct periods, the choice of comparable data is limited and thus the use of a simple parsimonious model makes sense. An added value of the econometric analysis is that the political system (the autocracy–democracy continuum and the political competition and participation, respectively) is included as an explanatory variable. The empirical investigation relates to the extent of deglobalisation during the period of import collapse of individual countries (the peak to trough movement). GDP per capita is not statistically significant: the econometric analysis does not find an impact from the level of development on deglobalisation. The manufacturing import share is significantly associated with stronger deglobalisation (this effect is strongest in the 1930s). The political system is highly significant in both 1930s and 2000s, but the impact is opposite. In the 1930s, autocratic rule and dictatorship are associated with stronger deglobalisation; in the 2000s, democracy is associated with stronger deglobalisation. The remainder of this article is organised as follows. The next section introduces and discusses some aspects of deglobalisation following a world trade collapse and motivates the selection of explanatory variables to be used in the small econometric model. The third section provides the detailed multivariate regression analysis that tests inter alia for differences in the impact of the variables in the 1930s viz. the 2000s. The final section draws conclusions and discusses some implications from the perspective of this Special Issue. Determinants of trade collapse and deglobalisation The study of deglobalisation has merit. Better understanding the causes and impact of collapsing world trade and deglobalisation is relevant for science because such real-world experiments provide important testing grounds for theory. The concurrence of declining imports and deglobalisation in a great many countries during and after the two world trade collapses implies that an export-led recovery is difficult for individual countries and for the world as a whole (van Bergeijk, 2009a). Studying deglobalisation builds the knowledge that is necessary for institutional design aimed at prevention and/or reduction of the impact of disruption of world trade. This article studies the peak to trough movement of the import to GDP ratio during the first 3 years of the trade collapse cum deglobalisation phenomenon. The reason to keep the analysis limited to a 3-year period is that the trade destruction during the Great Depression is concentrated in later years of the world trade collapse and occurred as a result of protectionism and inappropriate monetary policies (Kindleberger, 1978). Table 2 provides more details on the dynamics and patterns during the two periods, reporting the duration of the peak to trough movement of nominal and real trade flows.8 The trade collapse in the 2000s is characterised by high-speed dynamics. This stylised fact is confirmed for all four indicators (both real and nominal; exports and imports): trade flows hit bottom within 1 year in more than 80% of the available observations. This speed contrasts with the 1930s, where the mode and median are 3 years. Noteworthy is the homogeneity or agreement across indicators in the 2000s that contrasts with the heterogeneity or disagreement of the 1930s, especially regarding nominal and real developments, thus indicating that price movements (in particular deflation) were more important in the 1930s than in the 2000s. Table 2. Number of years between peak and trough of real and nominal trade in the 1930s and 2000s. 1 year 2 years 3 years 4 years Country observations 1928 Real exports 35% 21% 32% 12% 34 Real imports 15% 12% 62% 12% 34 Nominal exports 6% 74% 20% 35 Nominal imports 83% 17% 35 2008 Real exports 86% 11% 3% 158 Real imports 80% 14% 4% 1% 160 Nominal exports 92% 7% 1% 157 Nominal imports 89% 10% 1% 1% 150 1 year 2 years 3 years 4 years Country observations 1928 Real exports 35% 21% 32% 12% 34 Real imports 15% 12% 62% 12% 34 Nominal exports 6% 74% 20% 35 Nominal imports 83% 17% 35 2008 Real exports 86% 11% 3% 158 Real imports 80% 14% 4% 1% 160 Nominal exports 92% 7% 1% 157 Nominal imports 89% 10% 1% 1% 150 View Large The choice of the explanatory variables is based on both theoretical and practical considerations. The practical considerations are related to the availability of comparable data for the two periods concerned (actually the explanatory variables are measured in the year before the start of the trade collapse in order to avoid endogeneity issues that would otherwise plague the analysis). The theoretical model implicitly underlying the analysis is a basic international trade model under uncertainty. Increased awareness of trade uncertainty (due to the trade collapse) reduces international specialisation and thus the share of trade in domestic production. Behaviour under uncertainty (in particular risk-taking) is influenced by the extent of centralisation of decision-making and per capita income (van Bergeijk, 2009b, 47–65). Therefore, we will deal with institutions (political structures) and the level of development (per capita GDP). A third element is added to this set of explanatory variables in view of the significant attention in the literature for value chains and the composition effect of international trade (import share of manufacturing). Institutions Institutions are increasingly being recognised as an important determinant of international trade.9 Democratic countries tend to have more liberal trade policies (Milner and Kubota, 2005), which may reflect labour’s voting power (O’Rourke, 2006), but could possibly also reflect that ‘trade is less threatening to individuals who have confidence in their country’s political institutions’ (Mayda and Rodrik, 2005, 1410). van Marrewijk and van Bergeijk (1990, 1993) and Aidt and Gassebner (2010) pointed out that autocratic, centrally planned economies, due to their centralised decision-making processes, will respond quicker and sharper to (potential) trade problems. In the same vein, Eichengreen and Irwin (2009, 26) on the 1930s note: One might plausibly think that countries with authoritarian political regimes would be more likely to resort to exchange controls; restrictions on political freedom tended to go together in this as in other periods. In contrast, as for example pointed out by Rodrik (2017), the current problems appear to be created in a number of democracies that only a decade ago appeared to be strong and stable supporters of globalisation processes. Typically, the popular vote in some of the major democracies has turned against continued globalisation. However, the political organisation of the nation has not yet been considered in the empirical literature as a determinant of the size and impact of a world trade collapse. A related issue is the apparently shifting geopolitical role of the USA from hegemon to bully, which complicates the provision of global public goods (van Bergeijk, 2013a), including the multilateral trade and payment systems and the relevant global institutions.10 There is more at stake than the grumbling of former world powers that always accompanies the advent of a new hegemon (van Bergeijk and Moons, 2017). Regional economic integration initiatives have been cancelled or put into question (“open for renegotiation”), creating trade uncertainty and breaching the trust that is essential for engaging in international activities. The ensuing move towards isolationism and deglobalisation had its roots in the Great Recession. This clearly threatens the multilateral system in a slower but similar sense as the collapse of trade and international capital flows in the 1930s. Consequently, students of deglobalisation should specifically consider the political context and institutions. A second contribution of this article is that it provides a test of the hypothesis that democracy and autocracy matter for trade in the context of a world trade collapse and its aftermath, by including a political variable as an explanatory variable. Note that two different well-respected sources and definitions for the political institutions will be used to test robustness of the relationship to be estimated: Polity IV (autocracy–democracy continuum) and Vanhanen (political competition and participation). Level of development It is empirically important to consider the level of development, as illustrated by different patterns of the trade cycle for the advanced economies and the developing economies/emerging markets that can be observed both in the Great Depression (van Bergeijk, 2009a) and the Great Recession and its aftermath (van Bergeijk et al., 2017a). On the one hand, one might expect that trade at low levels of development consists of essential goods with low elasticity of demand so that reductions of imports would be expected to be more limited. On the other hand, financial leeway is often limited for developing countries so that they have to act more quickly and reduce imports if the prospects for export turn sour. In the theoretical model of uncertain trade implicitly underlying the quasi-postulated reduced form equation to be estimated below higher income is associated with a smaller reaction to an increase in the level of trade uncertainty. All in all it is important to include per capita GDP as a control variable also in view of significant welfare improvements between the 1930s and the 2000s. Manufacturing import share: international value chains and composition effects The major mainstream publications on the recent world trade collapse (Baldwin and Evenett, 2009; Freund, 2009; World Trade Organization, 2009) pointed out the importance of the composition effect (reductions of trade in durables were much stronger than in non-durables and services). A second item that has been investigated more in depth recently is the fragmentation of international production and its relation to trade intensity (Timmer et al., 2016). This is in line with the general recognition of the role of international value chains. The network of internationally interlinked firms could potentially explain both the propagation of shocks and the simultaneity in the timing of country-level trade collapses as well as severity of the trade shock (the fact that the percentage reduction of trade was a manifold of the percentage reduction in GDP). This value chain argument to a large extent rests on the observation that both the share of intermediate products in international trade and the elasticity of world trade to GDP increased significantly in the decades before the world trade collapse in 2008/2009. For example, Freund (2009) and Cheung and Guichard (2009) have proposed a causal relationship between the two observations and see fragmentation of production as a driver of the world trade collapse. The occurrence of value chain (and related composition) effects motivates the choice to use the manufacturing import share (for which data are available for both 1930s and 2000s) as one of the variables of interest in the econometric analysis in the next section. Since value chain activity typically is concentrated in manufacturing and involves intermediate products, this argument links to the earlier identified composition effect (less impact of the collapse on services, on non-durable consumer goods and on non-manufacturing goods). So Equation 1 that includes the manufacturing share in imports does not provide a test of these explanations separately but instead identifies the comparative strength of the role of manufacturing trade during the two periods of deglobalisation. The literature, importantly, does not support an unambiguous a priori expectation of the sign of this relationship. van Marrewijk (2009) argues that value chains spread the trade shocks over many countries providing a cushion. Brakman and van Marrewijk (2017) argue that involvement in international supply chains increases the susceptibility to (global) shocks but, at the same time, enhances faster recovery, and international value chains may thus also be a building block for resilience. Other examples of such resilience effects could be the larger trust regarding repeat buyers and the use of non-bank-intermediated trade credit (van Bergeijk, 2010). The sign of the relationship is thus not a priori clear and essentially an empirical matter. Empirical studies on the impact of international value chains during a trade collapse are contradictory. These studies use quite different approaches, including partial analyses (such as Tanaka, 2009; van Marrewijk, 2009), single-country analyses (such as Eaton et al., 2011; Levchenko et al., 2010; Robertson, 2009 with respect to the 1930s), calibrated simulation models (for example, Bénassy-Quéré et al., 2009; Bems et al., 2011; Eaton et al., 2011) and micro data (for example, Altomonte et al., 2012; Behrens et al., 2013; Tamminen, 2017; van den Berg and Jaarsma, 2017; Wagner and Gelübcke, 2014). van Bergeijk et al. (2017a) conclude that “the current debate on value chains is inconclusive or needs at least more nuance” (see also Gawande et al., 2015).11 Methodological differences could also perhaps explain contradictions. One important contribution of this article is that it provides a new data set that uses a multi-country multi-event perspective, and the first crude test of this hypothesis (by including manufacturing import share as an explanatory variable12) in a cross-country setting for the 1930s and the 2000s trade collapses and deglobalisation phases. Based on the discussion of the literature, the following quasi reduced form equation will be estimated. Change in globalisation=(α+δ1930sβ)political variable + (θ+δ1930sλ) Per capita GDP+ (γ+δ1930sξ) Manufacturing share+δ1930s+ C+ε (1) δ1930s is a shift and slope dummy that assumes the value 1 for observations regarding the 1930s and else zero, C is the constant term and ε is the error term. The inclusion of both additive and multiplicative (slope) dummies allows for the detection of different impact (size) during the Great Depression and the Great Recession for all explanatory variables (that are measured in the year before the trade collapse in order to reduce potential endogeneity problems). The appendix discusses the data collection and construction. Research design and econometric findings I analyse an unbalanced panel analysis consisting of 31 country observations during the Great Depression and 128 country observations during the Great Recession and its aftermath. The dependent variable is the percent change of import to GDP ratio, that is calculated on the basis of peak to trough movement of volume of imports and GDP between 1928 and 1932 and between 2008 and 2012, respectively.13 Table 3 reports the empirical findings. The difference between specification (1’) and (1”) is that specification (1’) uses the Polity data set and specification (1”) uses the Vanhanen data set. These data sets measure distinct concepts for the political institutions of a country and relate to slightly different country samples; the Vanhanen data set covers more small (island) economies. The different specifications are important to test the robustness of one of the major findings of this article, namely the impact of political institutions on the pace of deglobalisation. Table 3. Percentage change in globalisation (import to GDP ratio) 1929–1932 and 2008–2010. Specification 1’ 1” Number of observations 142 159 Source for political variable Polity (democracy–autocracy dimension) Vanhanen (political competition and participation) Political variable −0.66** (0.28) −0.31*** (0.11) Political variable × dummy1930s 1.8*** (0.60) 0.75** (0.36) GDP per capita −0.001 (0.001) −0.000 (0.000) GDP per capita × dummy1930s −0.006 (0.024) −0.000 (0.003) Manufacturing import share −0.18* (0.11) −0.11 (0.09) Manufacturing import share × dummy1930s −0.34** (0.17) −0.38** (0.17) Dummy1930s (shift dummy) −14.9 (13.7) −16.4 (−13.1) Constant term 8.5 (7.5) 5.4 (5.9) R2 0.31 0.29 Adjusted R2 0.28 0.25 F 8.7*** 8.6*** Specification 1’ 1” Number of observations 142 159 Source for political variable Polity (democracy–autocracy dimension) Vanhanen (political competition and participation) Political variable −0.66** (0.28) −0.31*** (0.11) Political variable × dummy1930s 1.8*** (0.60) 0.75** (0.36) GDP per capita −0.001 (0.001) −0.000 (0.000) GDP per capita × dummy1930s −0.006 (0.024) −0.000 (0.003) Manufacturing import share −0.18* (0.11) −0.11 (0.09) Manufacturing import share × dummy1930s −0.34** (0.17) −0.38** (0.17) Dummy1930s (shift dummy) −14.9 (13.7) −16.4 (−13.1) Constant term 8.5 (7.5) 5.4 (5.9) R2 0.31 0.29 Adjusted R2 0.28 0.25 F 8.7*** 8.6*** White heteroskedasticity-consistent standard errors in brackets. *90%, **95%, ***99% significance levels. View Large The regressions in Table 3 show that the two phases of deglobalisation are to a large extent comparable because the shift dummy for the 1930s is insignificant.14 More specifically, the results for GDP per capita (not significant in either period) and manufacturing import share (always negative, but stronger and more significant in the 1930s) are comparable in both specifications. A larger manufacturing import share is associated with stronger deglobalisation, especially during the Great Depression. The correlation between international value chain activity and globalisation (that constitutes the basis for the mainstream narrative on the impact of value chain activity on openness and international trade) requires a cautious interpretation, especially in view of the fact that the coefficient for manufacturing import share is larger (in absolute terms) and more significant for the 1930s, and this challenges the mainstream analysis. Globalisation is a firm-driven process and fragmentation of production according to the available evidence has been associated with an increase in the world’s trade to GDP ratio before the Great Recession. The underlying mechanism may, however, be quite different from the purely mechanical statistical relationship that relates to the different modes of measurement regarding GDP (value added) and trade (turnover). Value chain interaction may breed trust among participating firms because of the repeated-buy character of the transactions, and/or have external effects (such as demonstration effects, learning effects or network effects) which support globalisation. If this is the case, there is no reason why this role should be asymmetrical (positive in upswings and negative in downturns) as assumed by the mainstream analysis. This finding potentially has important implications both for the analysis and for policy advice, in particular the idea of stabilising economies by means of a reorientation towards domestic production. The findings also agree that political variables are a highly significant correlate of deglobalisation, but here an important difference between the two historical events is observed: in the 1930s, the coefficient is significantly positive, so that democratic countries are associated with less deglobalisation; in the 2000s, the coefficient is significantly negative, so that democratic countries are associated with stronger deglobalisation. Discussion This Special Issue reflects on the implications of the Great Recession for the neoliberal model that has been the basis of globalisation since the 1950s. The editors have specifically asked the question “Have we focused too much on ever more detailed, specific and locally orientated narratives at the expense of situating these adequately within more holistic accounts of large-scale systemic processes and structures?” This article answers this question in the affirmative, offering three contributions to the debate. First, the article provides an empirical analysis of the disruption of (neo)classical/liberal globalisation, identifying challenges to existing knowledges and approaches. This article thus provides an analysis of a specific empirical topic that is at the heart of this special issue. The empirical approach here is not grounded in a specific theory, to be useful for both heterodox and mainstream approaches, and it is definitely not meant to provide definite answers. The approach is especially versatile because it generates interesting research puzzles, for example: why was the composition effect (manufacturing import share) in the 1930s stronger than in the most recent deglobalisation phase? Second, the article argues and empirically shows that the analysis should not be limited to the current phase of instability and disruption. It demonstrates the usefulness of doubling the number of events regarding the phenomenon of deglobalisation in the wake of a world trade collapse during periods of economic disruption.15 It is from this perceived need for a more comprehensive analysis encouraging that new databases are being constructed, such as the RICardo project (Dedinger and Girard, 2016), TRADHIST (Fouquin and Hugot, 2016) and the History of Trade Agreements database (Kohl et al., 2017), that allow study of earlier global trade collapses (including the world trade collapses that occurred before the 1930s), to increase the scope for (more balanced) panels, and to include new explanatory variables (such as Free Trade Areas, Regional Economic Integration and Monetary Arrangements). This constitutes an important research agenda for economic geography, especially since the two world trade collapses studied here are characterised by significant geographical heterogeneity, implying that both analysis and policy need to avoid the one-size-fits-all trap of the early literature on the 2008/2009 trade collapse. Third, the empirical research establishes the need for a comprehensive and possibly multidisciplinary analysis that takes political factors and institutions into account in the analysis of deglobalisation. This finding suggests a change of the neoclassical (research) agenda that has mainly been focused on improving trade policies. In the current geopolitical and geo-economic context, the need, however, emerges to refocus research towards the political economy of protectionism, trade uncertainty, national economic security concerns (see van Bergeijk et al., 2017b) and the creation of physical barriers to trade and migration in the form of walls (van Bergeijk, 2015). The idea that politics needs to get a greater role in international trade analysis certainly is not new. Bhagwati (1991, xvi), for example, complains: ‘How can we possibly explain what happens unless we bring in the political equations into our modelling at the same time?’. The significant impact of political variables as one of the determinants of deglobalisation in the 1930s and 2000s is a robust and exciting result. The econometric findings, moreover, uncover new research puzzles: politics is a significant correlate for deglobalisation in the context of the two world trade collapses, but the sign of the impact of politics is opposite in the two periods. In the 1930s, democracies appear to reduce deglobalisation pressures. In the 2000s, deglobalisation pressures especially arise in democracies. Recent examples that corroborate this finding for the 2000s are the election of US President Trump, Brexit in the UK, The Netherlands referendum against the EU Treaty with Ukraine and the Belgium opposition against the trade agreement between EU and Canada.16 Are these only local incidents? Or is this the start of a new wave of protectionism and the prelude to a long period of deglobalisation? Acknowledgement This article is partially based on the updates of P.A.G. van Bergeijk, ‘次还不够！看待世界贸易崩溃和逆全球化的经济史视角’, International Social Science Journal (Chinese edition), Vol. 34, No. 1, pp. 17–30 March, 2017). References Aidt, T. S. and Gassebner, M. ( 2010) Do autocratic states trade less?, The World Bank Economic Review , 24: 38– 76. Google Scholar CrossRef Search ADS Altomonte, C., di Mauro, F., Ottaviano, G., Rungi, A. and Vicard, V. ( 2012) Global Value Chains During the Great Trade Collapse: A Bullwhip Effects? Working Paper Series No. 1412. Frankfurt am Main: European Central Bank. Aslam, A., Boz, E., Cerutti, E., et al.( 2017) The Slowdown in Global Trade: A Symptom of a Weak Recovery? Washington, DC: IMF. Available online at: http://www.imf.org/News/Seminars/Conferences/2017/06/21/~/media/57920FC394584433B676C2D47B3425E7.ashx [Accessed 5 August 2017]. Baldwin, R. E. ( 2009) Introduction: the great trade collapse: what caused it and what does it mean? In R. E. Baldwin (ed.) The Great Trade Collapse: Causes, Consequences and Prospects . London: Vox.EU.org and CEPR. Baldwin, R. E. and Evenett, S. J. ( 2009) The Collapse of Global Trade, Murky Protectionism, and the Crisis: Recommendations for the G20 . London: A Vox EU Publication. Behrens, K., Corcos, G. and Mion, G. ( 2013) Trade crisis? What trade crisis?, Review of Economics and Statistics , 95: 702– 709. Google Scholar CrossRef Search ADS Bems, R., Johnson, R. C. and Yi, K. M. ( 2011) Vertical linkages and the collapse of global trade, American Economic Review , 101: 308– 312. Google Scholar CrossRef Search ADS Bems, Rudolfs, Robert C. Johnson, and Kei-Mu Yi. The Great Trade Collapse . No. w18632. National Bureau of Economic Research, 2012. Bénassy-Quéré, A., Decreux, Y., Fontagné, L., et al.( 2009) Economic Crisis and Global Supply Chains , CEPII Document du travail 2009–15. Paris: CEPII. Bhagwati, J. N. ( 1991) Political Economy and International Economics . Cambridge, MA: MIT Press. Birnberg, T. B. and Resnick, S. A. ( 1975) Colonial Development: An Econometric Study . New Haven, CT: Yale University Press. Brakman, S. and van Marrewijk, C. ( 2017) Resilience, Supply Chains and the Great Recession , CESifo Working Paper Series No. 6505. Munich: CESifo. Campbell, D. L., Meissner, M. C., Novy, D., et al. ( 2009) Explaining two trade busts: output versus trade costs in the Great Depression and today. VOX , 19 September. Cheung, C. and Guichard, S. ( 2009) Understanding the World Trade Collapse , OECD Economics Department Working Paper No. 729. Paris: OECD. Chor, D. and Manova, K. ( 2012) Off the cliff and back? Credit conditions and international trade during the global financial crisis, Journal of International Economics , 87: 117– 133. Google Scholar CrossRef Search ADS Constantinescu, C., Mattoo, A. and Ruta, M. ( 2016) Does the global trade slowdown matter?, Journal of Policy Modeling , 38: 711– 722. Google Scholar CrossRef Search ADS Dedinger, B. and Girard, P. ( 2016) Exploring Trade Globalization in the Long Run: The RICardo Project. Available online at: http://ricardo.medialab.sciences-po.fr/Dedinger_Girard_RICardo_HistoricalMethods_revised.pdf. Dreher, A. ( 2006) Does globalization affect growth? Evidence from a new index of globalization, Applied Economics , 38: 1091– 1110. Google Scholar CrossRef Search ADS Eaton, J., Kortum, S., Neiman, B. and Romalis, J. ( 2011) Trade and the Global Recession , NBER Working Paper 16666. Cambridge, MA: NBER. Eichengreen, B. and Irwin, D. A. ( 2009) The Slide to Protectionism in the Great Depression: Who Succumbed and Why? , NBER Working Paper 15142. Cambridge, MA: NBER. Eichengreen, B. and O’Rourke, K. H. ( 2009) A tale of two depressions. VOX , 1 September. Available online at: http://www.voxeu.org. Estevadeordal, A., Frantz, B. and Taylor, A. M. ( 2003) The rise and fall of world trade, 1870−1939, Quarterly Journal of Economics , 118: 359− 407. Google Scholar CrossRef Search ADS Fouquin, M. and Hugot, J. ( 2016) Two Centuries of Bilateral Trade and Gravity Data: 1827–2014 , CEPII Working Paper 2016-14. Paris: CEPII. Freund, C. ( 2009) The Trade Response to Global Downturns: Historical Evidence , Policy Research Working Paper 5015. Washington, DC: World Bank. Google Scholar CrossRef Search ADS Gawande, K., Hoekman, B. and Cui, Y. ( 2015) Global supply chains and trade policy responses to the 2008 crisis, World Bank Economic Review , 29: 102– 128. Google Scholar CrossRef Search ADS Haddad, M., Harrison, A. and Hausman, C. ( 2010) Decomposing the Great Trade Collapse: Products, Prices, and Quantities in the 2008–2009 Crisis , NBER Working Paper No. 16253. Cambridge, MA: NBER. Haugh, D., Kopoin, A., Rusticelli, E., et al.( 2016) Cardiac Arrest or Dizzy Spell. Why Is World Trade So Weak and What can Policy Do About It? , OECD Economic Policy Papers 18. Paris: OECD. Hoekman, B. (ed.) ( 2015) The Global Trade Slowdown: A New Normal? CEPR. Available online at: http://www.voxeu.org/sites/default/files/file/Global%20Trade%20Slowdown_nocover.pdf. Hong, K., Lee, J.-W. and Tang, H. C. ( 2010) Crises in Asia: historical perspectives and implication, Journal of Asian Economics , 21: 265– 279. Google Scholar CrossRef Search ADS Ikeda, E. ( 2018) Global and Developing Country Business Cycles . PhD thesis, Erasmus University. International Monetary Fund (IMF)( 2016) World Economic Outlook (WEO) Subdued Demand: Symptoms and Remedies October . Washington, DC: IMF. Jacks, D. S., Meissner, C. M. and Novy, D. ( 2011) Trade booms, trade busts, and trade costs, Journal of International Economics , 83: 185– 201. Google Scholar CrossRef Search ADS Kindleberger, C. P. ( 1978) Manias, Panics, and Crashes: A History of Financial Crises . London and Basingstoke: MacMillan. Google Scholar CrossRef Search ADS Kohl, T., Milionis, P. and Mong, S. ( 2017) The History of Trade Agreements: A New Database . Groningen: University of Groningen. League of Nations( 1931) Statistical Yearbook of the League of Nations 1930/31 . Geneva: League of Nations Economic Intelligence Service. Levchenko, A. A., Lewis, L. T. and Tesar, L. L. ( 2010) The collapse of international trade during the 2008–09 crisis: in search of the smoking gun, IMF Economic Review , 58: 214– 253. Google Scholar CrossRef Search ADS Maddison, A. ( 1985) Two Crises: Latin America and Asia 1929−38 and 1973−83 . Paris: OECD. Maddison, A. ( 2006) The World Economy: Volume 1: A Millennial Perspective and Volume 2: Historical Statistics . Paris: OECD. Madsen, J. B. ( 2001) Trade barriers and the collapse of world trade during the Great Depression, Southern Economic Journal , 67: 848− 868. Google Scholar CrossRef Search ADS Marshall, M. G. ( 2011) Polity IV Project: Dataset Users’ Manual. Available at http://www.systemicpeace.org/inscr/p4manualv2010.pdf. Accessed on 1 December 2015. Mayda, A. M. and Rodrik, D. ( 2005) Why are some people (and countries) more protectionist than others?, European Economic Review , 49: 1393− 1430. Google Scholar CrossRef Search ADS Milner, H. V. and Kubota, K. ( 2005) Why the move to free trade? Democracy and trade policy in the developing countries, International Organization , 59: 107– 143. Google Scholar CrossRef Search ADS Murshed, S. M. ( 2017) Economic diplomacy and the liberal peace. In P. A. G. van Bergeijk and S. J. V. Moons (eds) Research Handbook of Economic Diplomacy , Chapter 16. Cheltenham: Edward Elgar. Nunn, N. and Trefler, D. ( 2014) Domestic institutions as a source of comparative advantage. In Handbook of International Economics , Vol. 4, pp. 263– 315. North Holland: Elsevier. OECD( 2009) Vertical Specialization and Global Value Chains , Working Party on International Trade in Goods and Trade in Services Statistics STD/SES/WPTGS(2009)16/ANN. Paris: OECD Statistics Directorate. O’Rourke, K. H. ( 2006) Democracy and Protectionism , IIIS Discussion Paper 191. Dublin: Institute for International Integration Studies. Paravisini, D., Rappoport, V., Schnabl, P., et al.( 2015) Dissecting the effect of credit supply on trade: evidence from matched credit-export data, Review of Economic Studies , 82: 333– 359. Google Scholar CrossRef Search ADS Robertson, R. ( 2009) Mexico and the great trade collapse. VOX , 27 November. Available online at: http://www.voxeu.org. Rodrik, D. ( 2017) Straight Talk on Trade: Ideas for a Sane World Economy . Princeton, NJ: Princeton University Press. Rothermund, D. ( 1996) The Global Impact of The Great Depression 1929−1939 . London and New York, NY: Routledge. Song, T., Lu, D., Liang, Y., et al.( 2017) Progress in international geopolitical research from 1996 to 2015, Journal of Geographical Sciences , 27: 497– 512. Google Scholar CrossRef Search ADS Tamminen, S. ( 2017) Regional effects or none? Firms’ profitability during the Great Recession in Finland, Papers in Regional Science , pp. 33-59. Tanaka, K. ( 2009) Trade collapse and vertical foreign direct investment. VOX , 7 May. Available online at: http://www.voxeu.org. Timmer, M. P., Los, B., and Stehrer, R. ( 2016) An Anatomy of the Global Trade Slowdown Based on the WIOD 2016 Release (No. GD-162) . Groningen: Groningen Growth and Development Centre, University of Groningen. United Nations ( 2010a) The 2008 International Trade Statistics Yearbook Volume I – Country Pages . Geneva and New York: UN. United Nations ( 2010b) The 2008 International Trade Statistics Yearbook Volume II – Trade by Commodity . Available online at: http://comtrade.un.org. United Nations Statistical Office( 1962) International Trade Statistics 1900–1960 . New York, NY: UN Statistical Office. van Bergeijk, P. A.G. ( 2009a) Expected extent and potential duration of the world import crunch, Kyklos , 62: 479− 487. Google Scholar CrossRef Search ADS van Bergeijk, P. A.G. ( 2009b) Economic Diplomacy and the Geography of International Trade . Cheltenham: Edward Elgar. Google Scholar CrossRef Search ADS van Bergeijk, P. A.G. ( 2010) On the Brink of Deglobalization: An Alternative Perspective on the World Trade Collapse . Cheltenham: Edward Elgar. van Bergeijk, P. A.G. ( 2013a) Earth Economics . Cheltenham: Edward Elgar. Google Scholar CrossRef Search ADS van Bergeijk, P. A.G. ( 2013b) The world trade collapse and international value chains: a cross-country perspective, International Economic Journal , 27: 41– 53. Google Scholar CrossRef Search ADS van Bergeijk, P. A.G. ( 2015) Visible and invisible walls: world trade patterns and the end of the Cold War, Acta Oeconomica , 65: 231– 247. Google Scholar CrossRef Search ADS van Bergeijk, P. A.G. ( 2017) Heterogeneity and geography of the world trade collapses of the 1930s and 2000s, Tijdschrift voor Economische en Sociale Geografie/Journal of Economic and Social Geography , doi: 10.1111/tesg.12265 (ahead of print). van Bergeijk, P. A.G., Brakman, S. and van Marrewijk, C. ( 2017a) Heterogeneous resilience and the Great Recession’s trade collapse, Papers in Regional Science , 96: 3– 12. Google Scholar CrossRef Search ADS van Bergeijk, P. A.G. and Mensink, N. W. ( 1997) Measuring globalization, Journal of World Trade , 31: 159– 168. van Bergeijk, P. A.G. and Moons, S. J. V. ( 2017) The rise of economic diplomacy research. In P. A. G. van Bergeijk and S. J. V. Moons (eds) Research Handbook of Economic Diplomacy , Chapter 1. Cheltenham: Edward Elgar. van Bergeijk, P. A.G., Moons, S. J. V. and Volpe Martincus, C. ( 2017b) The future of economic diplomacy research. In P. A. G. van Bergeijk and S. J. V. Moons (eds) Research Handbook of Economic Diplomacy , Chapter 23. Cheltenham: Edward Elgar. van den Berg, M. and Jaarsma, M. ( 2017) What drives heterogeneity in the resilience of trade: firm-specific versus regional characteristics, Papers in Regional Science , 96: 13– 32. Google Scholar CrossRef Search ADS van Marrewijk, C. ( 2009) Spatial Diffusion of Technology and the Trade Collapse . Utrecht: Utrecht University. van Marrewijk, C. and van Bergeijk, P. A.G. ( 1990) Trade uncertainty and specialization: social versus private planning, De Economist , 138: 15– 32. Google Scholar CrossRef Search ADS van Marrewijk, C. and van Bergeijk, P. A.G. ( 1993) Endogenous trade uncertainty: why countries may specialize against comparative advantage, Regional Science and Urban Economics , 23: 681– 694. Google Scholar CrossRef Search ADS Wagner, J. and Gelübcke, J. P. W. ( 2014) Risk or resilience? The role of trade integration and foreign ownership for the survival of German enterprises during the Crisis 2008–2010, Jahrbücher für Nationalökonomie und Statistik , 234: 757– 774. World Bank( 2015) What Lies Behind the Global Trade Slowdown, Global Economic Prospects . Washington, DC: World Bank. World Trade Organization( 2009) World Trade Report 2009 – Trade Policy Commitments and Contingency Measures . Geneva: WTO. NOTES 1 Bems et al. (2012, 29) conclude that “One benefit of the trade collapse is that it stimulated work on neglected areas at the intersection of international trade and macroeconomics.” 2 For the purpose of this article, a world trade collapse is defined as an event of negative annual real growth of international merchandise trade that occurs both in the aggregate for world trade and in the vast majority of countries for their individual imports and/or exports. I use the concept of world trade collapse to indicate the period of investigation, but the dependent variable in this article is the extent of deglobalisation that occurred after a world trade collapse. 3 A reduction of the trade flow in itself is neither a sufficient nor a necessary condition for a reduction trade openness as shown in my sample in the case of Japan, where GDP contracted stronger (faster) that imports. 4 An anonymous referee to the first version of this article. 5 See also World Bank (2015), Constantinescu et al. (2016), Haugh et al. (2016) and IMF (2016). 6 Recent studies that include the 1930s also by choice appear to neglect the 2008–2009 episode (for example, Jacks et al., 2011). 7 Also note that a third of the articles are single-country studies. 8 Current prices values are more frequently reported than real constant prices. 9 See Nunn and Trefler (2014) for a review of the literature on institutions and comparative advantage. 10 See Song et al. (2017) for an overview of recent geopolitical studies. 11 Empirical studies are contradictory. These studies use quite different approaches including partial analyses (such as Tanaka, 2009; van Marrewijk, 2009), single-country analyses (such as Eaton et al., 2011; Levchenko et al., 2010; Robertson, 2009 with respect to the 1930s), calibrated simulation models (for example, Bénassy-Quéré et al., 2009; Bems et al., 2011; Eaton et al., 2011) and micro data (for example, Altomonte et al., 2012; Behrens et al., 2013; Tamminen, 2017; van den Berg and Jaarsma, 2017; Wagner and Gelübcke, 2014). 12 See van Bergeijk (2013b) for a study on the role of value chains during the Great Recession that uses alternative measures of global value chain activity. 13 The choice of the import to GDP ratio as a dependent variable implies that one of the traditionally investigated determinants of the world trade collapse (GDP representing the demand shock of the Great Recession) drops out (I divide the left-hand and right-hand sides of the traditional approach by GDP). It is relevant to note that in a National Accounting framework an increase (decrease) of exports increases (decreases) GDP, while the opposite is true for imports. Hence, the choice for the import to GDP ratio avoids the problem of co-movement that could bias the use of the export to GDP ratio. 14 As noted before, this finding is corroborated for the world trade collapses in van Bergeijk (2017). 15 I have focused on the two major events, but other smaller episodes of world trade collapse cum deglobalisation could also be considered: for example, the years following the first oil crisis in 1973. 16 The election of French Prime Minister Macron apparently constitutes an important counterexample, but also in France, the increasing strength of the popular vote against globalisation needs to be recognised. Appendix The construction of the data sets is discussed in detail in van Bergeijk (2017). Import to GDP ratio is calculated using several sources: for the 1930s, United Nations Statistical Office (1962), Maddison (1985, 2006) and Birnberg and Resnick (1975) and for the 2000s, IMF World Economic Outlook Database (April 2015). The set of explanatory variables consist of country-specific variables. These are all measured in the year before the start of the world trade collapse that initiated the phase of deglobalisation. Manufacturing import share is derived and estimated from League of Nations (1931, Table 95, 168–170), United Nations Statistical Office (1962) and United Nations (2010a, 2010b). The political variables are taken from Polity IV data set (Marshall, 2011, see http://www.systemicpeace.org/polity/polity4.htm) and Vanhanen data set, 2011 (see http://www.prio.org/Data/Governance/Vanhanens-index-of-democracy/). Per capita GDP is derived from Maddison (2006), which is updated by the Groningen Growth and Development Centre (http://www.ggdc.net). © The Author(s) 2018. Published by Oxford University Press on behalf of the Cambridge Political Economy Society. All rights reserved. For permissions, please email: firstname.lastname@example.org
Cambridge Journal of Regions, Economy and Society – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud