Deglobalisation or decoupling: what four studies claim and measure

Ask two papers how globalised the world was at its peak around 2007–2008 and you get two numbers. Witt (2019) puts world imports of goods and services at 30.2 per cent of GDP and calls 2007 a peak (Witt, 2019). Ripsman (2021) puts merchandise trade at almost 51.5 per cent of world GDP in 2008 (Ripsman, 2021, p. 1322). Neither is wrong. They count different things, which is why one headline about "deglobalisation" can be right about one paper and meaningless about another.

This post reads four studies on the subject, so that when you meet "globalisation is dead" or "it is too late to decouple" you know what evidence sits behind it.

What does each of the four studies ask?

Each asks a different question and rests on a different kind of evidence.

The first two ask whether it has fallen. The third asks what follows if it has. The fourth asks whether separation is possible. Headlines tend to blur the three.

Has the world become less globalised? Witt and Chase-Dunn

Both papers measure trade and foreign investment against GDP, yet they land in different places. Witt finds the data consistent with deglobalisation. Chase-Dunn and colleagues say it is too soon to be sure.

Witt defines deglobalisation as weakening interdependence among nations, judged at world level. World imports of goods and services, averaged and weighted by GDP, peaked at 30.2 per cent of GDP in 2007, recovered to 30.1 per cent in 2010 and then edged down. Inward FDI flows peaked at 5.3 per cent of GDP in 2007 and have since stayed between 2 and 3 per cent (Witt, 2019). He hedges: currency conversion into dollars and inflation adjustment distort such numbers, so he reran them weighted by population and found the recent decline survived (Witt, 2019). Even inside his paper the peak depends on the series, from 2000–2001 (openness of investment rules) to 2011 (net FDI position) (Witt, 2019).

Chase-Dunn, Álvarez and Liao go further back. They use a series of imports as a share of national income in each country's own currency, from 1830 to 1995, which avoids converting to dollars, and set it beside World Bank data for 1970–2021, which is calculated differently. The series shows deglobalisation in 1879–1901 and 1921–1944, globalisation in 1945–1980, a plateau in 1981–1993 and a return to growth in 1994–2008. After 2009 they print a question mark (Chase-Dunn et al., 2023).

Their reason is duration. By their dating the two earlier deglobalisation phases lasted 22 and 23 years, and they have 13 years since 2008, so the prudent verdict is "too soon to tell". The similarities with earlier phases still lead them to lean towards deglobalisation or another plateau (Chase-Dunn et al., 2023).

One small lesson: Witt dates the FDI peak to 2007, Chase-Dunn to 2008. Witt accessed the World Bank FDI data in 2017, while the other team used 2023 data. I suspect data revisions explain the gap, but I have not checked. When two papers differ, ask when the numbers were pulled.

Does deglobalisation make the Great Powers more hostile? Ripsman

Ripsman does not ask whether trade has fallen. He takes deglobalisation as beginning in 2008 and speeding up from 2015, then asks whether it causes friction between the Great Powers (Ripsman, 2021, p. 1318). His answer is mostly no, and the causation may run the other way.

The argument rests on sequence. Cold War tension eased between 1986 and 1990 (Ripsman, 2021, p. 1320). Yet world trade as a share of GDP stayed in a narrow band, from just under 28 to just over 30 per cent, through 1983–1994, and then rose almost every year to nearly 51.5 per cent in 2008. FDI inflows also took off only from 1993 (Ripsman, 2021, pp. 1322-1323). His export series is in current US dollars, not adjusted for inflation, so read its direction rather than its size (Ripsman, 2021, p. 1322).

On structure, from 1985 Warsaw Pact countries took 42–49 per cent of their imports from their own bloc while producing under 20 per cent of world output. COCOM, in force from 1950 to 1994, barred Western countries from selling more than 100 categories of goods to the Soviet Union and its allies (Ripsman, 2021, p. 1324). He concludes that globalisation was largely a result of eased security tension.

He does not say the economy is irrelevant. If deglobalisation persists, he expects rivalry to shift from soft balancing towards harder tools (Ripsman, 2021, p. 1333). The weak point is that his trade and FDI data stop in 2008 and 2007; later deglobalisation is described through events such as Brexit and "America First" (Ripsman, 2021, p. 1329), not numbers.

What about decoupling? Farrell and Newman, and where the four meet

Farrell and Newman (2020) say it is too late to decouple. Their case is about structure, not trade levels: financial, supply-chain and internet networks converge on a few points of control, and states have learned to use those points as weapons (Farrell and Newman, 2020).

In April 2018 the US Treasury announced sanctions on Oleg Deripaska and his aluminium empire. The authors say Washington apparently failed to foresee the chaos in car and aircraft supply chains that relied on his products, so after lobbying by European firms and governments the sanctions were delayed and then unwound entirely. Defence procurement can be partly insulated, they argue, but remaking the consumer economy the same way would overturn whole industries and raise prices (Farrell and Newman, 2020). It is an essay: the cases were chosen because their consequences were large, and I do not know how typical they are.

Set side by side, three things stand out. First, neither Witt nor Chase-Dunn and colleagues blames technology. Witt writes that technology enables, while politics decides who can use it (Witt, 2019). Chase-Dunn and colleagues treat falling transport and communication costs as background, since costs rarely rose sharply when globalisation retreated (Chase-Dunn et al., 2023). If the cause is political, container freight rates are not the only signal worth watching; that is my inference.

Second, Witt (on his realist branch), Chase-Dunn and Ripsman all point to the relative weakening of the United States (Witt, 2019; Chase-Dunn et al., 2023; Ripsman, 2021, p. 1333). Third, the papers sketch different worlds ahead. Witt offers two: a patchwork, in which pairs or groups of countries keep different degrees of openness, or two blocs around the United States and China (Witt, 2019). Chase-Dunn and colleagues describe supply chains that are shorter but still cross borders (Chase-Dunn et al., 2023). Farrell and Newman call both total isolation and benign integration fantasies (Farrell and Newman, 2020). No study concludes that the world has already decoupled.

What should an exporter check after reading all four?

Swap "is globalisation over?" for "where am I dependent, and which scenario would hurt that point?" Each paper suggests one check.

Suppose you run a 60-person workshop in Haiphong assembling control units for industrial fans. It buys bare circuit boards from a supplier in Shenzhen and sells to a distributor in Poland and a customer in the United States. The workshop and the three partners are invented and do not represent real firms.

A note on the third row. Witt says operating in both blocs implies a risk of asset confiscation if rivalry turns into open hostility (Witt, 2019). The invented workshop has only sales contracts, so its risk is of another kind, one Witt does not analyse. To turn the two scenarios into decisions, the method in scenario planning for business owners will do. To test a trade chart before trusting it, use the five chart-reading questions.

This suits you if you have several customers and suppliers to compare. It does not suit you if nearly all revenue rests on one customer or nearly all inputs on one source: then the problem is dependence, not the macro scenario. That is my own inference, not checked against data on Vietnamese SMEs.

Going deeper: what did I read, and what not?

Witt (2019) is a perspective paper in open access. I read the full HTML text; figures and tables are images, so I relied on the descriptive text. This version has no print page numbers, so I cite without pages.

Chase-Dunn, Álvarez and Liao (2023) appeared in Social Sciences, also open access. I read the full HTML, did not open the supplementary files and could not read Tables 1–2. I did not read Bairoch either, so what the FAQ says about him is what the authors report.

For Ripsman (2021), the journal's site blocks automated access, so I read the publisher's full PDF through an Internet Archive copy; page numbers are print pages. Farrell and Newman (2020) appeared in Foreign Affairs, issue 1–2 of 2020, pp. 70–80. The essay has no DOI, so here is the publisher's page. I read the free online full text, which has no print pages.

One general limit: the data in these papers run from about 2015 to 2021, and I have not checked later figures, so the post cannot say where the world stands today. None of the four studies measures SMEs or Vietnam.