Discussion Paper

No. 2019-46 | August 12, 2019
Do remittances worsen export diversification?

Abstract

The paper explores the impact of workers’ remittances on the level of export diversification. The hypothesis is that significant inflow of remittances causes overvaluation of real exchange rate, which in turn deteriorates diversity of export. The theoretical base is in line with the Dutch disease phenomenon. The paper uses annual cross-national panel data over 2000–2016 period and System GMM methodology. The evidence suggests that indeed large inflow of remittances is associated with less diversified export. The economic intuition behind is that remittance-caused real exchange rate appreciation unevenly suppresses export of goods: some goods “suffer” more than others do. In terms of the number of product-names, a percentage point increase in remittances to GDP sent home “reduces” variety of export by approximately five active lines. There are other interesting findings as well. An improvement of government effectiveness facilitates overall export diversification; terms of trade improvement and rise of real exchange rate volatility mostly increase export concentration rather than alter number of exported product-names.

Data Set

JEL Classification:

F14, F24, F31

Assessment

  • Downloads: 667

Links

Cite As

Erik Vardanyan (2019). Do remittances worsen export diversification? Economics Discussion Papers, No 2019-46, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2019-46


Comments and Questions


Anonymous - Referee Report 1
August 15, 2019 - 08:25

Referee Report for “Do remittances worsen export diversification?”

The paper aims to show empirically that countries that receive remittances become less diversified, through the appreciation of the exchange rate channel. As the authors suggest, this is similarly/same as a Dutch Disease effect on diversification.

My main first comment ...[more]

... comes from that and relates to the contribution. The paper cited by the authors, Bahar and Santos (2018) exactly looks at that question: Dutch disease and export diversification. They find that indeed having an external shock (increase in price of natural resources in their case, but increase in remittance in this paper’s) results in less diversification. It is not clear to me what is then the contribution of this paper beyond that.

My second main comment is with respect to the interpretation of the coefficients, and particularly its economic significance. The coefficient of interest is the interaction between remittances and REER. But that’s quite confusing. That does not necessarily mean that remittances are creating a currency appreciation that is in turn causing lack of diversification. Those results suggest that when the two things are happening at the same time, is when you see less diversification; but not following the theory of change posed by the author. Moreover, the coefficients are quite small, even if statistically significant. I’m unsure how big these effects are. The author claims that a “percentage point increase in remittances to GDP sent home costs approx. five active lines”. Not sure how they arrived to that conclusion, but Column 6 of Table 2 (and 3) show a positive point estimate (not statistically significant) for active lines of that interaction term.

Lastly, though the authors claim that the GMM system solves the endogeneity problem, I’m not sure it does and it is not explained why is that. I think that it is ok to show correlations and acknowledge it, but it is less ok to claim they are causal effects.

The paper needs to be edited and rewritten. Often citations are in parenthesis at the beginning of a sentence, without years, or many typos and grammar issues.


Erik Vardanyan - Comments to the Referee Report for “Do remittances worsen export diversification?”
August 19, 2019 - 09:45

Thank you for your valuable Report.
You are right, Bahar and Santos (2018), as well as myself, are investigating the Dutch disease and export diversification relationship. In the first study, the disease-causing factor is the sizable export of natural resources, while for my case, this is the so-called 'export' of ...[more]

... the labor force. I believe, that those two factors have inherent economic differences. The economic impact of remittances may be different depending on the nature and characteristics of receivers. In other words, whether remittances are causing Dutch disease or not is an open question worth analyzing. In short, Bahar and Santos (2018)are asking to what extent the concentration of a country's non-resource export basket is determined by their exports of natural resources. While I am asking, to what extent the concentration of a country's overall basket is determined by remittances inflow they face. I believe, those questions sound similar but possess enough differences to be asked as separate research questions.
Regarding the second comment, I believe that it is not just a coefficient of the interaction term that we should look at. Primarily, I am interested in the coefficients of remittances and the interaction term altogether. One may look at the partial derivative of the export diversification equation with respect to remittances (the overall marginal effect of remittances).

Thank you.


Anonymous - Referee Report 2
December 05, 2019 - 08:22

see attached file


Anonymous - Comments
February 28, 2020 - 07:00

Thank you for your review.
Regarding the remittances and the real exchange rate relationship, for majority of cases investigated, appreciation channel prevails. Especially, for countries which are heavily dependent on this kind of financial means. But as I am mentioning in the paper, there are several country-specific studies finding depreciation ...[more]

... or no impact relationship at all. Though, comparing to the first group, those cases are rather uncommon. Anyway, it is important to consider this possibilities too. Notably, the main possible reasons of heterogeneity of the direction of the relationship could be the sample selection. By conducting the first regression analysis, I simply check whether my sample reflects this stylized fact. Here I am checking the sample, rather than improving the methodology, therefore the estimation approach is in line with existing literature. In other words, at this stage, I just need to be sure that my sample reflects the same appreciation relationship. Otherwise, the remittances – real exchange rate – export diversification story would not be consistent. In that instance, there will be no sound reason to continue research.
On the one hand, export diversification as a factor affecting the long run economic performance of an economy is an appealing hypothesis well discussed by Olivier Cadot, Céline Carrère and Vanessa Strauss-Kahn in “Export Diversification: What's behind the Hump?” 2011 and Jean Imbs, Romain Wacziar in “Stages of Diversification” 2003. On the other hand, Dany Bahar and Miguel A.Santos reviewed the Dutch disease phenomenon from the export concentration point of view. Those studies make pretty much the departure point for my own study. For remittances inflow dependent and other developing economies economic development and export diversification go hand-in-hand. Usually, those countries exports are represented by commodities and raw materials, so export diversification is one way to upgrades their technological base and pave new ways to other markets. Therefore, diversification affects economy of developing countries positively. I agree, in case of classical Dutch disease event the export concentration increase is a natural thing to observe and even the question may sound trivial. Not all sectors of export are going to develop as a booming one. Therefore, here the export diversification is going to be worsening. However, regarding the remittances, this logic is not applicable, there is no booming sector per se. All sectors of export are exposed to an appreciation shock equally, and the more competitive ones will suffer the least. I agree with the referee report, that it is not clear what sectors are more vulnerable and why. But this question is pretty much out of the paper scope. The most I am interested now is that exchange rate appreciation negatively affects total export and export diversification all together. The question about specific sectors may be investigated as another independent research.
The main differences between remittances and other types of financial flows is that the former one is usually aimed to be consumed. The beneficiaries of remittances are from relatively poor share of population with high propensity to consume. Therefore, higher remittances inflow usually directly translates into higher consumption. This is not necessarily true for example, for FDIs.
In terms of policy responses, as I stated in the paper, any restrictions on remittances inflow will eventually degrade population welfare. And the way out is not remittances restriction but rather moving away from remittances dependent state of the economy. Saving encouragement can be seen as measure alleviating negative consequences of remittances inflow, but newer as a measure to solve the problem.
Unfortunately, lack of microdata for developing countries does not allow addressing questions of underlying specific economic relationships. For instance, measuring size of income and substitution effects remittances. One way of explicitly showing those economic links is building up a structural model, but again this would be another research. Regarding the research methodology of the paper, I appreciate all comments and it would be great to hear some suggestions about alternative econometric techniques.