VoxEU Column Economic history International trade

The globalisation in the long run: Gains from trade and openness 1800-2014

The slowdown of global trade growth since the Global Crisis has raised concerns across the world. This column puts recent changes into perspective by presenting evidence on the export/GDP ratio and a rough measure of the gains from trade back to 1830. It shows that the interwar period was marked by a reversal of globalisation that makes recent trends look like a small blip. 

The stagnation of world trade since the outbreak of the Global Crisis in a very long term perspective has been framed before (Baldwin 2009, Hoekman 2015), including in our previous column (Federico and Tena-Jungito 2016a). We have concluded that “the effect of the Great Recession on long term growth of trade is sizeable but so far fairly small in comparison with the joint effects of the two World wars and the Great Depression”. To be sure, the world GDP has continued to grow, although at a substantially lower rate – 4.2% versus 8.2% in the previous seven years (UNCTADstat), while it plummeted during the Great Depression. Thus, in this column we net out the effect of GDP changes by focusing on movements in export/GDP ratio (Federico and Tena-Jungito 2016).

Export/GDP ratio since 1830: New findings

We have estimated openness before 1938 by dividing exports at current prices from our data base (Federico and Tena-Jungito 2016b) by the series of GDP from the available historical reconstructions, and after 1950 with UN data. We have been able to compute the ratio for 38 countries at the peak of the two globalisations, in 1913 and 2007. Twenty-nine of them were more open in 2007 than in 1913, and the ‘world’ ratio was about 80% higher (12.5% in 1913 and 22.5% in 2007). The sample is highly representative, including a majority of third world countries, and it accounted for 90% of world trade in 1913 and 75% in 2007. Anyway, the world would have been more open in 2007 than in 1913 even if all missing polities had exported nothing. Unfortunately, data for quite a few of these countries are missing in some years. We thus present yearly series since 1830 for 17 countries (‘1830 sample’) and since 1870 for 27 ones (‘1870 sample’), which we plot in Figure 1 alongside the export/GDP ratio for all countries after 1970.

Figure 1. World openness, 1830-2014

Sources: Federico and Tena-Junguito (2016b)

A visual inspection suggests a division in three periods, with two major waves of globalisation separated by a century of fluctuations:

  • The export/GDP ratio for the ‘1830 sample’ more than doubled, from about 6% in 1830 to almost 14% in 1870 - i.e. it grew as fast as export/GDP for all countries from 1972 to 2007;  
  • The export/GDP ratio for the ‘1870 sample’ sample grew very little if not at all from 1870 to 1913, fell during the World War I, remained low in the 1920s, collapsed to a minimum of 6.7% in 1936, and recovered only partially during the Golden Age.

As a result, the aggregate ratio in the early 1970s was similar to the level of the late 1920s and about two points lower than a century before.  

  • Openness rose very fast from the early 1970s to 2007, doubling (from 9.9% to 19.6%) for the ‘1870 sample’ and increasing by 2.5 times (from 9.3%  to 25.6%) for all countries.

This latter ratio soared during the oil crises, stagnated in the 1980s and early 1990s, and grew fast again from 1995 to an all-time peak in 2008. It is too early to assess the most recent trends. The ratio fell sharply in 2009, but recovered in the following year and anyway the decline is not comparable to the 30% collapse during the Great Depression. On the other hand, there is no sign of recovery of the pre-crisis upward trend.

Exploring the causes of growth of openness

One might find the growth of openness since the 1980s somewhat at odds with the standard narrative of the globalisation triumphant. We explore the proximate causes of the (too slow) growth of openness by decomposing it among changes in distribution of GDP by country (location effect), changes in the share of tradeables on GDP of each country (structural change), and a residual, which should capture the effect of change in trade costs (as well as, admittedly, all errors of estimation). This latter account for all the growth in the long run – i.e. from 1870 to 2007 (for the ‘1870 sample’); without changes in location and in composition of GDP, the export/GDP ratio would have risen by 13.4 points, to 24.6%. The residual was positive before 1913 and after 1973, while it was heavily negative in the interwar years (1913 to 1950) because of the widespread adoption of protectionist and autarkic policies during the Great Depression. Yet, even in those years, the location effect reduced openness almost as much as protectionism. The rise of (relatively closed) US on total GDP alone accounts for three quarters of the total decline in export/GDP. During the second globalisation, the location effect was negligible, and the growth of openness was dampened by structural change. Without it, the export/GDP ratio in 2007 would have been a third higher.

Last but not least, we have estimated the static gains from trade according to the sufficient statistics by Arkolakis et al. (2012) for 36 countries. As expected, gains were higher in 2007 in 25 of them, with an average 11.5% (median 7.9%) than in 1913 (average 6.3% and 5.4%, respectively). We obtain a yearly series of world gains by weighing the polity series with their shares on total GDP of the 1870 sample at current prices. As expected, given the formula (essentially the ratio import/GDP raised to the power of 3.78, as estimate of trade elasticity), the movements shadow those of export/GDP (see Figure 2).

Figure 2. Gains from trade 1830-2007, GDP weighed

Source: Federico and Tena-Junguito (2016b)

The aggregation of country series gives much weight to advanced countries and indeed an alternative, population-weighted series for the 1870 sample yields a somewhat different picture (see Figure 3).

Figure 3. Gains from trade 1870-2007, different weighing 

Source: Federico and Tena-Junguito (2016a)

The difference between the two series is mainly due to India, which accounted for 5% of GDP and half of the population of the sample. India was arguably an extreme case of de-globalisation in the 1950s and 1960s, but the experience of rich and poor countries differed markedly throughout the whole period.

  • The first globalisation before 1870 benefitted rich countries more than poor ones, which in the 1830 sample were represented by Latin American and European peripheral countries.

After 1870, the gains of the advanced countries fluctuated between 4% and 4.5%, while those of the rest of the world, including the Ottoman Empire and India, caught up. The Great Depression hit both groups badly. The gains for the rich plummeted below the level of 1830 and increased very slowly during the Golden Age, while those of the rest of the world recovered in the late 1930s and during the war, but declined in the 1950s and 1960s. As a result, gains in the early 1970s were lower than in 1913 by two fifths for the advanced countries and by one fifth in the rest of the world.

  • The second globalisation benefitted significantly both groups but gains increased a bit faster in rich than in poor countries.

An increase by 8% of GDP relative to no trade, although not negligible, may seem at odds with the hype about the benefits of globalisation. However, subsequent research has shown that the baseline Arkolakis et al. formula undervalues the true gains, suggesting alternative and more realistic formulas, which however need data unavailable in historical perspective (Costinot and Rodriguez-Clare 2014, Felbermayr et al. 2015). Furthermore, no static estimate can capture the dynamic effects of globalisation.


Our historical analysis suggests that, contrary to a widely held view, the current level of globalisation of trade is unprecedented. Openness and gains from trade at their peak in 2007 were substantially higher than in 1913 and the gap is substantially greater if one considers only tradables. Furthermore, the second globalisation should be compared to the period 1830-1870 rather than to 1870-1913. These latter years featured widely different trends by country, which yield a long-term stagnation. Of course, it is too early to tell whether this is an harbinger of future trends, or the upward path of globalisation will resume.


Arkolakis C, A Costinot, and A Rodriguez-Clare (2012) ‘New Trade Models, Same Old Gains?’ American Economic Review  102: 94–130

Baldwin R (2009), The Great Trade Collapse: Causes, Consequences and Prospects, A VoxEU.org Publication Geneve and London Graduate Institute and CEPR

Costinot A and A Rodriguez-Clare (2014), ‘Trade theory with numbers: quantifying the consequences of globalization’ NBER WP 18896 [Chapter in Gopinath, Helpman and Rogoff (eds) Handbook of international economics Vol 4 Elsevier-North Holland].

Felbermayr G, B Jung and M Larch (2015), “The welfare consequences of import tariffs: a quantitative perspective”, Journal of International Economics, 97: 295-309.

Federico, G and Tena-Junguito, A (2016a),"World trade, 1800-2015", VoxEU.org, 7 February 2016.

Federico, G and Tena-Junguito, A (2016b), 'A tale of two globalizations: gains from trade and openness 1800-2010', CEPR DP11128.  

Federico G and A Tena-Junguito (2016c), “World trade, 1800-1938: a new data-set,” EHES Working paper N93 (click here for the dataset).

Hoekman, B(2015), The great trade slowdown: a new normal? CEPR eBook.

4,094 Reads