Remittances and Business Cycles: Comparison of South Asian Countries

ABSTRACT

South Asia is one of the world's principal remittance-receiving regions. This study examines the home and host business cycles of migrant remittance flows to the region. Employing the Structural Vector Autoregression (SVAR) technique, the remittance behaviour of the region's four main countries is compared. Remittances to India and Pakistan show a mainly acyclical behaviour with respect to the output of the four host regions, and a countercyclical behaviour with respect to home output. In contrast, remittances to the two smaller economies of Bangladesh and Sri Lanka are found to be mainly procyclical. The study shows that the macroeconomic remittance behaviour varies with respect to the importance of remittance flows in the home economy. Moreover, remittance behaviour seems to respond more to home economy specificities than to those of the different regions that host the migrants from the developing countries.

Keywords:

• remittances
• business cycles
• South Asia
• Bangladesh
• India
• Pakistan
• Sri Lanka
• SVAR

JEL Classifications:

• E32
• F15
• F22
• F24

Notes

¹We separated the host economies into four major geographic regions, namely North America (consisting of the USA and Canada), the Gulf Cooperation Council (GCC) (comprising Saudi Arabia, the United Arab Emirates, Kuwait, Qatar, Bahrain and Oman), Europe (the United Kingdom, Germany, France, Italy, the Netherlands, Spain, Greece, Belgium, Ireland, Switzerland, Sweden, Denmark and Norway), and Asia Pacific (consisting of Japan, Australia, Singapore, Hong Kong, Malaysia and New Zealand).

²A brief description of the FD filter is given in Appendix B.

³A brief description of the critical values for the statistical significance of the cross-correlation coefficient is given in Appendix C.

⁴Appropriate lag length is determined using the Akaike, Schwarz, Hannan-Quinn and Final Prediction Error criteria. The model with the lowest value for the criteria is selected.

⁵The ‘identification problem’ calls for imposing restrictions on some of the structural parameters. The identification through the Cholesky decomposition is considered a mechanical technique that some believe is unrelated to the economic theory.

⁶The main problem in the structural model estimation is that one cannot directly estimate the variables of interest such as A and B₁ in equation (5) (Gottschalk, 2001)

⁷The cross-correlations were also obtained using a maximum of three leads and lags, but the results were found to be identical.

⁸The correlation that is statistically significant and reported to the left of contemporaneous correlation is accidental and can be ignored. This result would imply that either remittances to Pakistan or the country's output leads the host output.