- International Monetary Fund. European Dept.
- Published Date:
- October 2011
(a) Constructing a Trade Interconnectedness Index
The trade interconnectedness index (TII) is the weighted average of indicators capturing the importance of bilateral trade between countries for a group of N countries. The weight reflects the closeness of trade relationships.
Specifically, the TII for country i is
The importanceij indicator is composed of nine different criteria, eight of which measure the importance of imports and exports relative to an individual country’s economy as well as to that of trading partners. These include:
1. Mij/Yi and Mij/Yj
2. Xij/Yi and Xji/Yj
where Mij (Xij) is imports (exports) of country i from (to) country j; Mi (Xi) is imports (exports) of country i; Yi is GDP of country i; and
These individual indicators take the value of 1 (otherwise 0) when the underlying measurement exceeds the specified threshold, which is set at the 75th percentile for each criterion.
For the ninth criterion, the indicator is based on the size of bilateral trade: (Mij + Xij) relative to trade of all countries. For this indicator, the threshold is set at the 90th percentile to capture only substantial bilateral trade pairs.
The closenessij measurement is based on the notion of how directly countries are connected through trade. The construction of this measurement consists of two steps.
The first step is to specify what could be considered an important trade linkage. The analysis takes the view that a bilateral trade linkage between country i and country j is important if the importanceij indicator takes the values of at least
The second step is to count the distance between country i and country j, which in turn provides the value of closenessy,ij being defined as the inverse of the shortest distance. For example, if two countries share an important bilateral trade linkage and are thus connected directly, the shortest distance is 1 and the closeness is also 1. In contrast, if two countries are connected only through another country that has important trade linkages to both, the shortest distance is 2 and the closeness is ½.
Once both importanceij and closenessij are constructed, the TII can be computed. This can be done for any specific group of countries. For instance, when the degree of trade interconnectedness of Europe as a whole is of interest, the TII is calculated based on all European countries. On the other hand, when the degree of trade interconnectedness with Central Europe is of interest, the TII is calculated with respect to countries in Central Europe (that is, computing the weighted average of importanceij where j represents all countries in Central Europe).
(b) Quantifying Output Spillovers through the Trade Channel
The analysis aims at quantifying the magnitude of output spillovers through the trade channel based on the structure of bilateral trade relationships within the region.
The analysis relies on two key assumptions. One is that the export multiplier (that is, the magnitude of output change due to export change) is equal to 1; thus, the analysis does not account for additional output spillovers within the economy, leakages of domestic demand to imports, and intermediate imports essential for production. Another is that additional output spillovers across countries are not considered.
The analysis estimates the percentage change in output in country i owing to a 1 percent change in output in country j which is denoted by
Based on the assumption of unitary export multiplier, the change in output in country i results from the change in exports from country i to country j (owing to a change in output in country j), which is, in fact, imports from country i by country j. Than,
The analysis calculates the values of si|j for countries in Europe as well as in the Asia and Pacific region. Figure 4.14 presents the magnitude of output spillovers as a result of a 1 percent increase in output in all other countries in each region. This is simply the sum of si|j over all countries j.
(c) Growth Spillovers in a VAR Framework
The growth spillovers between Western Europe (WE) and CESEE are examined using a standard VAR framework containing quarterly real GDP growth for the sample period of 1997:Q2–2011:Q1, controlling for growth shocks that originated in the rest of the world (ROW). All major countries outside Europe are included in the VAR model to make sure that the estimated impulse responses purely reflect the spillovers between Western Europe and CESEE rather than reflecting similar responses to common global shocks.
For the purposes of this analysis, ROW includes the United States, emerging Asia (China, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Philippines, Singapore, Taiwan Province of China, and Thailand), Japan and all other economies in the IMF’s global projection model (Argentina, Australia, Brazil, Canada, Chile, Colombia, Israel, Mexico, New Zealand, Peru, South Africa, and Venezuela). PPP-based weights are used to construct the aggregate growth rates of ROW.
The following VAR models are estimated:
VAR-I: [ROW, WE, CESEE], and
VAR-II: [ROW, WE, CE, CESEE excluding CE].
The VAR-I system is estimated to study the spillovers between Western Europe and CESEE (Figure 4.15, left panels). The VAR-II model serves to examine the growth linkages between Western Europe and Central Europe (CE) on the one hand, and Western Europe and rest of CESEE, on the other hand (Figure 4.15, right panels).
The identification of the estimated shocks is achieved using Cholesky decompositions (that is, standard recursive ordering), and results presented in the text used the ordering of the countries indicated above. Robustness analysis for the result employing the methodology proposed by Bayoumi and Swiston (2008) was also carried out. More specifically, alternative orderings among countries were considered, and “averaged” impulse responses were calculated. The results are not affected by alternative orderings.
The model is estimated with five lags to ensure absence of autocorrelation in the estimated residuals. The results with four lags, which are more standard in the literature using quarterly data in the estimation of VAR models, yield quantitatively similar results.
(d) Credit Spillovers
The role of western bank lending to CESEE in the credit boom-bust cycles (Panel VAR-I) and the relationship between real credit growth and real economic activity growth (real GDP growth, real domestic demand growth, and real import growth) in CESEE (Panel VAR-II) are studied using a Panel VAR approach.
The baseline scenario considers 15 CESEE countries (Belarus, Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, the Slovak Republic, Slovenia, Turkey, and Ukraine) and estimates the Panel VAR model using the least square dummy variable (LSDV) method between the 2003:Q1–2010:Q4 periods.58
This model aims at characterizing the relationship between western bank lending and private credit expansion in CESEE. The first stage Panel VAR model, estimated with country-specific dummies using two lags, is as follows (for i = country index and t = 2003:Q1–2010:Q4):
where CYi,t is the quarterly change in private credit relative to GDP, and FYi,t is the quarterly change of BIS-reporting banks’ exposure to CESEE banks relative to GDP.
The second step of the analysis studies the relationship between real credit growth and real economic activity.
The second stage Panel VAR model, estimated with country-specific dummies using two lags, is as follows (for i = country index and t = 2003:Q1–2010:Q4):
where Zi,t is quarterly growth of real economic activity, and Ci,t is quarterly growth of real private credit.