Abstract
This study examines the determinants of worker's remittances. Variance decompositions, impulse response functions and Granger causality tests derived from a vector error correction model are used to test if remittances are affected by the macroeconomic conditions of the host (remittance sending) or home (remittance receiving) country. Data from Brazil, Colombia, the Dominican Republic, El Salvador, Mexico and the US are used. The results indicate that remittances respond more to changes in the macroeconomic conditions of the host country, than to changes in the macroeconomic conditions of the home country.
Acknowledgements
We would like to thank Susan Pozo, Mark Wheeler and Isabel Ruiz for helpful comments and suggestions. Any remaining errors are ours.
Notes
1 In this case household refers to the emigrant's family in his country of origin. Host country is the country to which the individual emigrated. Home country is the emigrant's country of origin.
2 Other papers testing the determinants of remittances using microeconomic level data include Agarwal & Horowitz (Citation2002), Brown (Citation1994 Citation1997), de la Brière et al. (Citation2002), Hoddinott (Citation1994) and Merkle & Zimmerman (Citation1992).
3 For example Amuedo-Dorantes & Pozo (Citation2004) found that remittances appreciate the real exchange rate of the receiving countries.
4 In this case U 1 is the derivative of utility with respect to home country consumption.
5
for a maximum.
6 Because of data limitations the receiving country's GDP is used only for the Mexican case and not for the ROW case. Interest rate differentials were included in the estimation but were not found to be significant.
7 All the series are seasonally adjusted and were tested for seasonal unit roots.
8 We also constructed a quarterly series of US remittances outflows interpolating annual US remittances outflows using quarterly US net aggregate remittances. The main results did not change. See Russell (Citation1986) for more data issues related to remittance measures.
9 Dissemination Standards Bulletin Board, Mexico Summary Methodology for the Balance of Payments, at www.imf.org.
10 The only exception is the quarter that includes 11 September 2001. We conducted our estimations excluding this quarter from our data and the results did not change.
11 CPIMEX is Mexico's CPI, CPIBRA is Brazil's CPI, CPICOL is Colombia's CPI, CPISAL is El Salvador's CPI and CPIDR is the Dominican Republic's CPI.
12 The Cholesky decomposition ordering is remittances, US FFR, US M2, US CPI, US unemployment, and the home country variable. For robustness we also estimated the model with the home country variable first in the ordering, there was not a big differences in the results.
13 See Engle & Granger (Citation1987) for more details on the estimation of vector error correction models.