The Spaghetti Bowl: Visualising the Proliferation of Preferential Trade Agreements


The world of preferential trade agreements (PTAs) is rapidly evolving: governments increasingly negotiate on the bilateral and regional level. This activity has created a complex web of PTAs spanning the globe. Economics professor Jagdish Baghwati likened the phenomenon to a “spaghetti bowl” in the early 1990s. Since then, new waves of preferential trade negotiations have swept over different regions, leading to a considerable increase in the overall number of treaties and the network’s density.

The figure above (Click here for a larger version) depicts the whole network of PTAs (a total of 910 treaties, including accessions to base treaties) for 203 countries/customs territories. It is based on the DESTA dataset, which I have visualised using a combination of R and Gephi.

How to read the figure:

  • Countries are depicted as circles, treaties are represented by rectangles; the size of the individual nodes is proportional to their respective number of connections.
  • Lines connect a country with a treaty: individually, they represent membership of a specific country in a PTA.
  • The colours depict geographical regions. Interestingly, the mapping algorithms automatically create regional clusters, which highlights the fact that many governments negotiated the majority of their agreements with states in their immediate neighbourhood.

A few months ago, I also created a dynamic version to visualise these treaty connections.

In comparison, this is how the network looked like in 1995 – when the World Trade Organization (WTO) was established and also approximately when Bhagwati made his observation of the spaghetti bowl: (Click here for the larger version)


The figure above shows all treaties in the DESTA dataset that were concluded before 1996; note that the colouration differs slightly from figure 1.

The Spaghetti Bowl: Visualising the Proliferation of Preferential Trade Agreements

A Brief Look at Diversity in the German Parliament

In light of the upcoming German federal elections in September 2017, I take a brief look at the age and gender distribution in the German Parliament – the Bundestag – from its establishment in 1949 to the last general election in 2013. The figure below sums up the most interesting findings, starting with the observation that while the age composition changed over time, the average age of the representatives  stayed relatiely constant over the past seven decades (at around 50 years at the start of the legislative term). In addition, it highlights that the number of female representatives has risen considerably from a meagre 28 out of 382 members in the 1st Bundestag (1949-1953) to 230 out of 631 in the current (18th) term.


Continue reading “A Brief Look at Diversity in the German Parliament”

A Brief Look at Diversity in the German Parliament

Global Trends in Regime Development and Democratisation

I recently worked with a few fascinating datasets that describe the transformation and democratisation of national governments over the last decades. Based on this very comprehensive data, this post discusses important trends in global regime development from 1972 to 2015 and also provides a few more detailed graphs for Africa, which is the focus of my current research for the German Institute of Global and Area Studies (GIGA). See here for some of my previous work at GIGA.

As the data below shows, the global political landscape has changed significantly over the last decades. A large majority of countries made considerable progress in liberalising their political systems and establishing democratic institutions. Regularly scheduled and increasingly competitive elections have become the norm and most countries are now governed by constitutions that are – at least on paper – more or less democratic. However, authoritarian rule has persisted or reappeared in many regions of the world and this has led to different trajectories in regime development. Some seasoned observers even argue that democratic progress has slowed significantly and that the last wave of democratisation might now be succeeded by a long phase of stagnation and decline.


Source: Own calculation based on the variable ‘regime1ny’ in Wahman, Teorell, and Hadenius (2013); see their codebook (and below) for more information on the data. 

The figure above summarises the changes in the type of government for a large majority of countries between 1972 and 2010. In particular, it shows the slow but steady increase in the number of democracies, highlights the sudden post-Cold War transformation from one-party to multiparty regimes and hints towards the stagnation of democratic progress over the last decade.

Comparing this with ‘raw’ data from the Freedom House Index (FHI, on which the categorisation in Wahman, Teorell, and Hadenius (2013) is partly based), we can identify similar trends when it comes to both civil liberties and political rights. In the FHI, a country is assigned two ratings each year – one for political rights and one for civil liberties. Each is rated between 1 and 7, with 1 representing the greatest degree of freedom (!) and 7 the smallest degree of freedom. Below you find what is called a ‘violin plot’ that summarises the development in both ratings for the FHI. The figure shows the distribution of cases from 1975 to 2015 in five-year intervals. The black diamonds depict the average score in the given year across all countries; the horizontal black lines describe the quantiles.

We can clearly identify a significant strengthening of citizens’ civil liberties and political rights. For example, in 1975 more than half of all countries were rated 5 or worse in political rights and only one-quarter received a score of 3 or better, but this situation reversed dramatically: in 2015, only about one-quarter received a rating of 5 or worse and almost half of all countries were rated 3 or better (one-quarter was even rated 2 or better). The average in both rankings also decreased by more than one point between 1975 and 2005 (also between 1975 and 2015). More recently, averages slightly increased again, but at least the median for political rights in 2015 shows that this might primarily be due to some extreme cases rather than an overall retreat of democracy.


Source: Own calculation based on FHI data (1975-2015)

Of course, reducing the very broad spectrum of real-world government types to a few (more or less distinct) regime categories simplifies the issue and many interesting questions remain unanswered. One of these questions is, for example, whether ‘multiparty regimes’ – which technically still are authoritarian regimes – have become more democratic since 1990. I try to provide an initial answer to this question below by looking at the size of the majority that the governing party enjoys in the country’s legislative assembly. This offers a relatively good indication of the level of political contestation (openness of the political system and electoral competitiveness) in the country as well as of the strength of the domestic opposition. A lot has been written on the emergence of ‘electoral authoritarian regimes‘ or ‘competitive authoritarianism‘, but this post is not a (good) literature review.

[Also see below for figures on regime development in Africa.]

Continue reading “Global Trends in Regime Development and Democratisation”

Global Trends in Regime Development and Democratisation

The Economic Impact of Sanctions against Russia on EU Member States


While attending a workshop on international sanctions a few weeks ago, I was fortunate to meet the author of the fascinating and very recent article “The Redistributive Impact of Restrictive Measures on EU Members: Winners and Losers from Imposing Sanctions on Russia“. Much has been written on EU-Russia relations following the crisis over Ukraine and important obstacles to a rapprochement remain; however, there have been few other detailed assessments of the economic impact of the sanctions, countersanctions and geo-political uncertainty on EU member states. Because the magnitude of the overall economic effects was a surprise to me – trade contracted by over a third for almost all EU member states and by much more for some members since 2013 – I couldn’t resist to take a brief look at the changes in the EU-Russia trade relationship myself: my analysis below complements the cited work but is based on more fine-grained data, which offers a potentially more nuanced assessment of some aspects of the development in EU-Russia trade relations.

“The concerns expressed by several EU leaders regarding the cost of the restrictive measures imposed on Russia were justified.” (p. 15)

In response to the annexation of Crimea by the Russian Federation in March 2014 and Russia’s support of armed separatist forces in eastern Ukraine, a group of states, led by the European Union and the United States, has imposed separate but overlapping sanctions on Russian individuals and businesses. EU sanctions have repeatedly been broadened in scope and today include restrictions against proponents and beneficiaries of Russian actions in Ukraine, economic sanctions against state-owned banks, energy and defence companies, as well as limitations on economic exchanges with Crimea. On 13 March 2017, the European Council prolonged the restrictive measures for a further six months, until 15 September 2017.

Sanctions are an important part of the policy response to what the EU and the US consider illegal actions by the Russian government. Germany has particularly emphasised the importance of a political solution and has strenuously worked towards a negotiated settlement: in March 2015, EU leaders decided to align the existing economic sanctions to the complete implementation of the Minsk II agreement, a package of measures to de-escalate the military confrontation in the Donbas that Germany and France facilitated between Moscow and Kiev. The provision of economic aid and military capabilities to Ukraine has been another pillar of the Western response; Germany has mostly focused on financial contributions.

This brief assessment aims at quantifying the changes in the EU-Russia trade relationship – which reflect economic sanctions and countersanctions as well as other factors, such as investor uncertainty and considerable changes in the exchange rate. The article also provides evidence on which member states have been affected the most in their trade relations with Russia and whether there are countries or sectors with excessive export losses or gains. This becomes especially relevant with regard to present discussions over burden-sharing and demands for a compensatory mechanism on the European level, as well as the Russian countersanctions on agricultural products, which may place disproportionately higher costs on the Southern and Eastern European countries.

In Sections 1 and 2, the text provides some background on the rationale behind the introduction of economic sanctions against Russia. This will then be complemented in Section 3 by a quantitative analysis of EU-Russia trade flows since 2013.

Continue reading “The Economic Impact of Sanctions against Russia on EU Member States”

The Economic Impact of Sanctions against Russia on EU Member States

An Example of the Successful Use of Export Taxes and Its Value for North-South Trade Negotiations

A recurring problem in the discussion on North-South trade relations is identifying good examples that show how controversial – trade-distorting – policy instruments are successfully used to promote economic development. The search is not a purely academic exercise as the case studies can be used to legitimise and defend policy tools in trade negotiations aimed at outlawing or restricting their domestic application. Export duties, i.e., taxes imposed upon the export of raw materials, are one of these instruments and the Economic Partnership Agreements (EPAs) between the European Union and developing countries are one attempt at circumscribing their use.

Because export taxes can be used to ensure a price advantage to domestic industries and therefore skew international competition – a good example are Chinese duties on the export of rare earth minerals – or enrich a small authoritarian elite, their use is actively discouraged by many industrialised countries. However, export taxes can also incentivise producers/exporters to process raw materials domestically into higher-value goods or components (that can be exported without additional charges). If used correctly, export taxes can thus promote the economic development and industrialisation of developing countries. Existing WTO rules do not discipline Members’ application of export taxes and only few countries have agreed to binding constraints on the use of export taxes during their WTO accession.

The EPAs between the EU and regional groups made up of African, Carribean and Pacific countries would limit the ability of governments to use export taxes considerably. For example, Article 13.1 of the EPA between the EU and the Economic Community of West African States (ECOWAS) declares: “No new duties or taxes on exports or charges with equivalent effect shall be introduced, nor shall those currently applied in trade between the Parties be increased from the date of entry into force of this Agreement.” According to Article 13.3, African members can impose export charges only “in exceptional circumstances, on a temporary basis and after consulting the European Union Party […] and with equivalent effect” of existing export charges.

But can export taxes be effective in promoting economic development? And if so, are there good examples that should discourage us from restricting their use? A brief look at the development of the Ethiopian leather industry suggests some benefits from the use of export taxes:

In February 2008, Ethiopia introduced a 150 percent tax on the export of raw and semi-processed animal hides and skins. This was meant as an instrument to encourage industries engaged in the preparation of raw hides and skins for export to shift to more advanced processing stages. Consequently, exports in raw hides and skins dropped significantly in 2009 and remained low, but exports in processed goods (“tanned or crust hides and skins”) almost doubled until 2011. In 2012, the Ethiopian government added a further 150 percent tax on the export of crust leather, i.e., leather that has been tanned, dyed and dried, but not finished. Again, this resulted in a signficant drop in exports of the affected products and the transformation of the leather industry to perform more advanced tasks in country.

The figure below summarises this development. It is based on data collected by the International Trade Centre for product group 41 (raw & semi-raw leather) and product group 42 (manufactured leather products). In Figure 1, I aggregate the different product types into four categories: (i) “raw hides” describes the most basic products, i.e., raw hides and skins (HS 4-digit: 4101-4103); (ii) “tanned hides” constitutes tanned or crust hides and skins” (HS 4-digit: 4104-4106); (iii) “prepared leather” represents more advanced leather processing (HS 4-digit: 4107-4113); and (iv) “manufactured leather products” are all finished leather products in product group 42.

Figure 1 – Ethiopian leather exports to all trade partners


Source: own calculation based on ITC data

Overall, the most drastic changes in the distribution of exports seem to occur in close temporal connection with the introduction of export taxes, as indicated by the two dashed vertical lines. The quick second transformation of exports in tanned hides to prepared leather from 2011 to 2012 suggests that the creation of this tax was better communicated and affected industries anticipated the costs. In addition, the adoption of this final processing step might have been much less demanding than the initial transformation. In parallel, exports in manufactured (finished) leather products started to increase from 2011. The same effects are visible in the trade relationship with the EU, see Figure 2.

These findings suggest that export taxes were used effectively to transform the Ethiopian leather sector from an industry focused on the preparation of raw skins to more advanced processing stages, while increasing the overall value of exports and encouraging the production of finished products. Of course, it is likely that other factors such as the growth in external demand, foreign investment and other policy interventions affected the transformation. The magnitude of the effects and temporal connection nevertheless suggest a considerable (positive) effect of the exports taxes on the economic development of the Ethiopian leather industry.

Trade agreements that are too restrictive of this and similar policy instruments might thus undermine national development strategies in the long run. Thus, it will be crucial for all members to the EPAs, the EU and its partners, to actively use review clauses such as Article 13.4 of the ECOWAS-EPA, which allow for regular reality-checks and revisions to the agreements “taking full account of their impact on the development and diversification of the economy of the West Africa Party” or other developing partners.

Figure 2 – Ethiopian leather exports to the European Union


Source: own calculation based on ITC data

A brief addition:

Below I complement my assessment of the impact of export taxes on Ethiopian leather exports by also including exports of shoes with a leather component. These products are included in product category 64 (footwear), so I select all shoes with some leather content based on the HS 6-digit level. (This equals all product lines in category 6403, as well as line 640420 and 640510.)

I find that exports of leather shoes broadly mirror the growth rate in exports of other more advanced leather products. Ethiopian exports in shoes with leather content consequently increased specifically between 2011 and 2013, i.e., the time period where the government introduced export taxes on processed leather. This correlation, which also can be observed for the exports of other finished leather products, suggests that export taxes may have encouraged the domestic production of more advanced leather products.

Figure 3 – Ethiopian leather exports to all trade partners (incl. leather shoes)


Source: own calculation based on ITC data

Ethiopia also recovered from the slight decrease in the exports of leather shoes after 2013 in 2016, when exports more than doubled to about $39 million.

An Example of the Successful Use of Export Taxes and Its Value for North-South Trade Negotiations

The UK-EU Trade Relationship: Part II

Looking again at UK trade data, I discuss what tariff costs UK exporters would face in the unlikely scenario that no Brexit deal (or interim trade agreement) can be reached and commercial relations would have to continue on a ‘most favoured nation’ (MFN) basis. The analysis is based on product groups established in the Harmonized System (HS) on the two-digit level, which allows to calculate some broad estimates of potential costs, but does not constitute a detailed assessment for each traded product.

1. An Update on UK Exports in 2016

Using data provided by the International Trade Centre (based on Eurostat data), I complement my previous analysis in Part I with information on UK exports in 2016. Interestingly, the data shows that UK exports to the EU increased again relative to 2015 while trade with other regions decreased. The difference between both trade flows narrowed from nearly £38,000 million in 2015 to about £17,000 million in 2016. See below for a brief overview:


Source: own compilation based on ITC & Eurostat data

2. UK Exports to EU Members by Product Group

As I have already shown in my first post, new markets for UK goods emerged rapidly over the last decade and British companies have increasingly found new customers in East Asia and other regions. Yet, a significant amount of UK exports are still directed towards Europe. For three-quarters of all HS-2 product groups, a majority of exports were traded with the EU in 2016 – see below for the quantiles:

0% 25% 50% 75% 100%
2.91 49.95 60.91 70.84 92.62

The following figure summarises these findings for each product category: the x-axis depicts the UK trade volume with Europe in percent; the y-axis shows the number of HS-2 product groups for each segment. The dashed line depicts the average UK trade across product categories with the EU, which equals 58.94 percent.


Source: own compilation based on Eurostat data

3. EU MFN Tariffs 

The big question for the immediate future is whether the EU and the UK will be able to spell out an agreement that preserves most aspects of current relations concerning trade in goods. A free trade agreement (FTAs) would, at a minimum, consist of abolishing tariffs on most or all goods traded between the parties, thus leaving both sides free to conclude FTAs with other countries – which is an explicit goal of the UK government.

Preserving the mutual access to national markets that EU member states currently enjoy certainly constitutes a priority for both sides in the Brexit negotiations. The diversity of interests within both parties and diverging priorities – as well as any conditionality on other parts of the package Brexit deal – could, however, render the negotiations on trade in goods laborious and time-consuming. (Of course, the negotiations on trade will also be affected by many more issues, such as product standards, rules of origin, rules on the services trade, taxation, dumping, …)

In case no (interim) agreement is reached in time, the UK would revert back to WTO rules, i.e., UK exports to the EU would be subject to the EU’s MFN tariffs. The figures below describe the current EU MFN tariff structure. They both show that EU MFN tariffs are actually relatively low, with an average (dashed lines) between 4.7 (figure on the left) and 6.6 percent (on the right), and only few extreme values for individual HS-2 product groups.


Source: own compilation based on Eurostat data 

Of course, there is a technical detail that makes both figures important for different reasons: in the one on the left, the average tariff for each product group is calculated by dividing the sum of tariffs through the number of all products in the group (including those without a tariff). This reduces the average tariff and introduces groups with an average tariff of practically zero. On the right, the sum of tariffs in a particular HS-2 product group is only divided by the number of dutiable items in the same group, thereby creating a potentially more accurate representation of the true impact of the EU tariff on traded goods, but reducing the utility of the figure for an assessment which does not distinguish between specific products and, instead, merely looks at the product groups as a whole.

4. Potential MFN Tariff Impact on UK Exports 

What would be the immediate costs for UK exports to the EU if MFN-tariffs would apply and which sectors could be affected most?

In 2016, the UK exported goods with a total value of £142,244 million to the EU. Multiplying the value of UK exports in each product category with the respective EU MFN tariff yields tariff costs of approximately £5,140 million, i.e., an effective tariff rate of 3.61 percent. Caveat: Because I calculate the potential tariff costs for whole HS two-digit product groups and tariffs vary widely for individual products, my results can only present a very rough estimate of potential costs.

Overall, there seems to be a linear positive relationship between the trade volume and tariff costs, with only a few outliers. Interestingly, product category 87 “Vehicles and parts thereof” would incur particularly high tariff costs (about £1,126 million) – a finding that is consistent with the present debate about the future of the UK automotive industry after Brexit. The 15 product groups with the largest trade volumes to the EU could potentially face the following tariffs:

HS Product Group HS Group Name UK Exports to EU 2016 in £ million Potential Tariff Cost in £ million
22 Beverages, spirits and vinegar 2742.44 32.18
27 Mineral fuels/oils 13111.46 190.47
29 Organic chemicals 3963.96 166.42
33 Essential oils; perfumery, cosmetic or toilet products 2643.90 62.71
38 Miscellaneous chemical products 2631.10 137.02
39 Plastics and articles thereof 5603.66 310.11
61 Apparel and clothing, knitted or crocheted 2036.47 236.30
62 Apparel and clothing, not knitted or crocheted 2469.59 285.57
71 Pearls, precious stones, precious metals 2686.91 15.59
72 Iron and steel 2065.63 4.83
84 Machinery and parts thereof 16241.82 311.27
85 Electrical machinery and equipment 10058.70 304.49
87 Vehicles other than railway and parts 17884.69 1125.57
88 Aircraft, spacecraft, and parts thereof 7246.19 222.56
90 Precision Instruments 5083.13 102.23

The figure below briefly summarizes the statistics for all product categories. The UK exports to the EU for each HS-2 product category are plotted on the y-axis; the potential tariff costs on the y-axis.


Source: own compilation based on ITC & Eurostat data 

5. Tariff Cost vs. Export Dependency on Europe

Finally, are UK exports highly dependent on the EU as a market and does this affect some product categories with high potential tariffs?

To answer this question, I subdivide the product categories into four groups based on the percentage of goods in each category that are exported to the EU. (See the table describing quantiles above.) I find that the first quantile – the product groups for which the trade dependence on the EU market is relatively low, i.e., below 50 percent of goods go to EU members – account for about £2,092 million of potential tariffs. This, includes several product categories with large trade volumes, for example, 87 for motor vehicles and 84 for machinery.

The product categories in quantiles two and three represent £779 million and £1,104 million respectively. The HS-2 product groups with the highest dependence on the EU as an export market, which describes categories in which more than 70 percent of exports go to EU member states, would incur approximately £1,164 million in tariff costs.

The figure below provides a rough overview of potential costs by plotting the EU’s MFN-tariffs on the x-axis and the UK exports to the EU for each HS-2 product category on the y-axis. All observations are labelled with the two-digit number of the respective HS product group. In addition, the colour of the individual observations indicates the amount of UK exports in the respective product group that is traded with EU member states.


Source: own compilation; note that one observation with an extreme tariff value of 45 percent (product group 24 for tobacco products) was omitted in the figure.


The UK-EU Trade Relationship: Part II