| | Intercept | Δ Portfolio ret (-1) | Δ Portfolio ret (-2) | Δ Portfolio ret (-3) | Δ SP500 (-1) | Δ SP500 (-2) | Δ SP500 (-3) | Δ Brent Oil (-1) | Δ Brent Oil (-2) | Δ Brent Oil (-3) | Dummy GFC | Dummy NGFC | ECT (-1) |
|---|
Latin America | 7.4263*** (1.9230) | 0.0588 (0.0634) | 0.0139 (0.0644) | 0.0637 (0.0620) | − 0.0336 (0.1323) | − 0.0840 (0.1386) | − 0.1957 (0.1342) | − 0.3906 (0.2957) | − 0.6019** (0.3163) | 0.3076 (0.2683) | – | – | − 0.0055 (0.0048) |
Asia | 1.9367*** (0.3361) | − 0.0002 (0.0019) | 0.0021 (0.0019) | − 0.0010 (0.0019) | 0.7585*** (0.0221) | 0.0831*** (0.0223) | 0.0234 (0.0220) | − 0.0547 (0.0509) | − 0.0071 (0.0539) | – | – | – | – |
MENA | − 0.8131*** (0.1170) | − 0.0012*** (0.0005) | 0.0010** (0.0005) | − 0.0001 (0.0005) | 0.5397*** (0.0079) | 0.0158** (0.0081) | 0.0178** (0.0078) | − 0.0465*** (0.0164) | − 0.0066 (0.0165) | – | – | – | – |
CEE | 2.2903* (1.3375) | − 0.0108** (0.0147) | 0.0003 (0.9572) | 0.0008 (0.0047) | 2.3326*** (0.0855) | 0.2791*** (0.0867) | 0.0988 (0.0850) | 0.1265 (0.2093) | 0.7026*** (0.2303) | 0.3924** (0.2075) | – | – | – |
- This table presents a VECM model for four emerging market groups. We only present model 3: \({\Delta P}_{i,r,t}={\delta }_{2i}+ {\theta }_{1i}{\sum }_{k=1}^{n}\Delta {P}_{j,r-i,t-k }+ {\theta }_{2i}{{\sum }_{k=1}^{n}{\Delta BRENT}_{it-k}+}{\theta }_{3i}{\sum }_{k=1}^{n}\Delta {S\&P500}_{it-k} {+{\delta }_{1i}{ECT}_{it-1}+\epsilon }_{it.}\) For the other models see Additional file 1: Table S5-1 to S5-4. Here, \({\Delta P}_{i,r}\) is a vector of returns of emerging markets, i, in region r; \({\Delta P}_{j,r-i,t}\) (columns 4–6) is a vector of returns of j markets in the yellow portfolios, excluding i. \({Brent}_{it}\) is Brent oil price series; and \({S\&P500}_{it}\) is the price index of the US market. These variables appear in first differenced form, represented by \(\Delta\). \(\delta\) and \(\theta s\) are the parameters to be estimated. The error correction term (ECT), which is one lag of the residual from Eq. (1) if significant and negative, confirms a stable long-term relationship between the variables identified. The lag structure for the model is chosen by minimizing the Schwarz Information Criteria. Values in parenthesis are standard errors. *, **. And *** denotes level of significance at 10%, 5% and 1%, respectively
