Sequencing in Customs Union Formation: Theory and Application to Eurasian Economic Union Gerald Pech...

Sequencing in Customs Union Formation: Theory and Application

to Eurasian Economic UnionGerald Pech

based on joint work with Aidos Alimbekov and Eldar Madumarov

Sequencing as a Problem

• Why do economic unions not form in one go?• Historical coincidence? – Eastern Europe• Evolving nature of union?• Increasing attractiveness over time

– could bring in a “strategic” motive behind the sequencing

A “Strategic” Explanation of Sequencing

• Aghion-Antras-Helpman, J Internat Econ (2007): “equilibrium sequencing”

• An “agenda setter/formateur” can form union at once or start with a core union

• Assume that forming the core union has a “negative externality” on an accession candidate

• It should be willing to join under conditions more favourable to the formateur

• Prediction: The country which is most negatively affected joins last

Kyrgyzstan’s delayed entry

• Signatory of the 1995 agreement – but not of the Dushanbe agreement 2007

• WTO member since 1998• April 2010: Bakiyev steps down• Atambaev wins election on promise of close

ties with Russia• Kyrgyzstan joins in 2015

Effect of Customs Union on Kyrgyzstan

• Re-import business and textile industry historically benefitted from low tariff regime

• Kyrgyzstan expected to be a net-loser from joining customs union

• But Kyrgyzstan also suffers negative external effects from core customs union

• Main industries equally hit by exclusion from integrating market

How to Explain Sequencing in Eurasian Customs Union

• So by joining Kyrgyzstan loses compared to status quo ante

• But once core customs union has formed joining is the least bad option

• So is the strategic sequencing model the best explanation of what happened with Kyrgyzstan’s entry?

A Historical Precedent: The Zollverein

• The German Customs Union (Zollverein) was founded in 1833 between independent German states

• Development that started at Prussia’s initiative in 1816

• And finally resulted in German nationhood

www.geraldpech.net 7

The Zollverein: Hesse-Cassel

• In 1828, Hesse-Darmstadt joined Zollverein• Negative externalities on Hesse-Cassel

– through higher border-tariffs of neighbouring Darmstadt

– resulting in civil unrest– and fear of further loss of transit traffic

• Recalcitrant duke resigned in 1831, successor signed ratification documents

The Zollverein

www.geraldpech.net 9

www.geraldpech.net 10

An Aside: Why Was Prussia Interested in Integration?

• Prussia did not benefit that much from trading with the smaller states

• But it a gained better bargaining position when negotiating over trade with big players such as England and France

www.geraldpech.net 11

What is missing from the story

• As Alimbekov/Madumarov/Pech (2015) show, the AAH story either needs – myopic accession candidates– an exogenous constraint on bargaining institutions

which the agenda setter can choose from – bargaining frictions: it takes time to bargain

• Kyrgyzstan’s WTO membership makes negotiations more complicated

• Problems in negotiating with Bakiyev gvt’– better modelled as “political uncertainty?”

Richer Explanations

Some more facts

• Kyrgyzstan receives Russian infrastructure support

… some more facts

• New border facilities at Kazakh-Kyrgyz border– Making core customs union more credible and

costly?• Russia promises Kyrgyzstan help in securing its

outer border – suggests that desire to “incentivize” Kyrgyzstan to

invest into border controls could have played a role

An Imperfect Commitment Model of Accession

• Assume accession is irreversible• and accession candidate cannot credibly

promise to implement policies which are necessary to make union successful

• Like investing in border security• Sequencing solves the incentive problem:

Make accession conditional on investment

Supporting Materials

www.geraldpech.net 17

Winners and Losers

• Economic welfare of a country is sum of– Consumer surplus– Tariff revenue– Profits of home producers– We assume that foreign firms have market power!

• As a proxy for impact of CU we use– change in current account balance – change consumer surplus of non numeraire good

• Political benefits and costs!www.geraldpech.net 18

Eurasian Customs Union

• 1995: First agreement signed between Belarus, Russia, Kazakhstan, Kyrgyzstan, Tajikistan and Uzbekistan.– Uzbekistan not interested at the moment– Tajikistan only has land border with Kyrgyzstan

• 2007: Dushanbe agreement on customs union signed between BY, KZ and RU

• 2010: Official launch of CU with external tariff at (high) Russian rates– for some goods July 2010, for others July 2011

www.geraldpech.net 19

Tariff Protection

simple average 2009

weighted average 2009

Kazakhstan 6.78% 5.52%

Belarus 8.00% 2.3%

Russia 8.09% 5.9%

EU 10.6% 7.37%

from: Jandasov/Sabyrova 2009

www.geraldpech.net 20

Standard and non standard channels

Source: Vashakmadze, E., Kaminski, B., Mironova, Y., (2011).

www.geraldpech.net 21

Standard and non standard channels

www.geraldpech.net 22

• • where is the average value of the export index for country i

in the pre- and the post customs union period (t=1,2), CU is a dummy variable for customs union which takes the value 1 for the subsample starting in 2010, CDi is the country fixed effect and Ii is a dummy which is 1 if the country is a latecomer in the customs union (i.e. Tajikistan (TAD), Kyrgyzstan (KY) or Armenia(ARM)) and 0 if the country as of now is not expected to join the customs union (i.e. Ukraine, Azerbaijan, Moldova).

0 1 72,..,6it s i is

EXP CU CD CU I

Dependent Variable: Method: Least Squares Date: 05/03/15 Time: 17:39 Sample: 1 12 Included observations: 12 =b0+ b1*CU+ b2*TAD+ b3*KY+ b4*ARM+ b5*MOL+ b6*AZ+ b7*(KY+TAD+ARM)*CU

Coefficient Std. Error t-Statistic Prob.

b0 238.1018 96.99498 2.454785 0.0701b1 289.1980 96.99498 2.981577 0.0407b2 -166.0115 137.1716 -1.210246 0.2928b3 0.672807 137.1716 0.004905 0.9963b4 49.87698 137.1716 0.363610 0.7346b5 -179.2669 118.7941 -1.509056 0.2058b6 383.8645 118.7941 3.231343 0.0319b7 -235.6465 137.1716 -1.717896 0.1609

R-squared 0.921567 Mean dependent var 338.6452Adjusted R-squared 0.784308 S.D. dependent var 255.7868S.E. of regression 118.7941 Akaike info criterion 12.62738Sum squared resid 56448.15 Schwarz criterion 12.95065Log likelihood -67.76429 Hannan-Quinn criter. 12.50770F-statistic 6.714100 Durbin-Watson stat 2.690579Prob(F-statistic) 0.042389

itEXP

itEXP