Do remittances cause Dutch Disease? A meta-analytic review

ABSTRACT

This is the first meta-study to analyze the effect size of remittances on the real exchange rate (RER). Using 426 estimates from 67 studies, we found that remittances appreciate RER; however, the effect size differs across countries. Compared to global estimates, the mean effect of remittances on RER is significantly larger for East Asia and the Pacific, whereas the same nexus is found to be significantly smaller for Latin America and the Caribbean. Furthermore, the synthesized effect of remittances on RER is found to be significantly positive for lower-middle-income countries. Our empirical examination shows that publication selection bias is less likely in the literature. We found a genuine effect of remittances on RER appreciation. This study also provides policy recommendations and presents suggestions for future research in the field.

KEYWORDS:

• Remittances
• real exchange rate
• Bayesian model averaging
• Frequentist model averaging
• weighted-Average least squares

JEL CLASSIFICATION:

• C4
• F24
• F31
• F40
• F41

Acknowledgment

This research was financially supported by Cape Breton University grant number RISE81116. We thank Tom D. Stanley for his helpful comments and suggestions on our earlier results. We also thank Asif Aslam Bholat and Arshia Khalid for their literature search, and Cindy Butler for her editorial support. All errors are our own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Within our included studies, no paper has used cross-sectional data.

2 The World Bank (2021b) country classification criterion was used to categorize countries by regions and income groups.

3 Four studies provide nine useable estimates that include an interaction between remittances and another variable.

4 For equations and mathematimatical details, see Cazachevici et al. (2020).

5 We employed the bms package available in R developed by Zeugner and Feldkircher (2015).

6 We are thankful to an anonymous reviewer for suggesting this estimation technique.

7 A PRISMA diagram of literature selection for meta-analysis is shown in Figure A1.

8 As explained above in Section II, a positive (negative) PCC value yields appreciation (depreciation) of RER.

9 We tested publication bias for the short- and long-run separately (results are not shown here, but available upon request). The results exhibit that PSB is unlikely for both cases.

10 The cutoff values determining a small effect, medium effect, and large effect are 0.10, 0.30, and 0.50, respectively (Cohen 1988). Nevertheless, Doucouliagos (2011) argued that this criterion is somewhat strict in economics literature that employs many control variables. Therefore, Doucouliagos (2011) proposed 0.104, 0.226, and 0.386 as the lowest threshold values for small, medium, and large effects, respectively, as the new general standard in macroeconomic research. For this study, we use Doucouliagos (2011) threshold values as a criterion to measure the strength of the effect size.

11 Kass and Raftery (1995) and Havranek and Sokolova (2020) proposed that if the value of PIP is equal to or greater than 0.5 in BMA, we may determine a non-negligible impact on the dependent variable.

12 Due to the lack of a sufficiently large sample, we are unable to examine the remittances-RER nexus for the short-run dataset separately.

Additional information

Funding

The work was supported by the Cape Breton University [RISE81116].