The effect of remittances prior to an election

¹ Nevertheless, a part of remittances could be indirectly seized by the governments through the broadening of the value-added tax (VAT) base.

² Faye and Niehaus (Citation2012) demonstrate that foreign aid is provided to more aligned countries during election years. This argument supports the idea that aid will promote more political manipulation during competitive elections.

³ To deal with the well-known problem of instrument proliferation raised by the system-GMM estimator (Roodman, Citation2009), the matrix of instruments is collapsed and the number of lags is always set at a level which ensures that the number of instruments does not exceed the number of cross-sections.

⁴ As we will show later in the article, the government final consumption variable appears to be the main fiscal variable sensitive to electoral manipulation during the election year, compared to other forms of fiscal outcomes (e.g. public investment).

⁵ The GDP per capita of migrant host countries is computed as the weighted sum of GDP per capita of all potential destination countries, with weights being the bilateral migration shares extracted from the Global Bilateral Migration Database published by the World Bank. For example, the instrument for remittance inflows into Mali is computed as the weighted sum of GDP per capita of all potential destinations of Malian migrants abroad, with weights being the proportion of Malian migrants living in each of those destinations at each period of time. Conditional on the presence of the control variables (including some other channels of globalization and domestic income), the instrument is assumed exogenous.

⁶ Imposing such a restriction could be misleading, insofar as the in-sample remittance data do not exactly include 0. The minimum value of the remittance-to-GDP ratio across the specifications is 0.004% of GDP.

⁷ Another approach might consist of an estimation using the Caner and Hansen (Citation2004) methodology but with the assumption that the threshold variable is exogenous in a nondynamic model. However, these restrictions seem very difficult to apply in our context.

⁸ The SE of the threshold level is computed using a delta method, that is, by taking a first-order Taylor approximation around the mean.

⁹ This definition of endogeneity is not perfect. It will consider the election after the death of the political leader as endogenous, while it is not in the sense of not being strategically chosen by the incumbent government. We are aware of this important drawback, but we do not have a way of correcting for it. We rest therefore on the strategy proposed by the literature.

¹⁰ The World Bank maintains one of the most reliable and often-used databases on remittances. Within the data broadly considered, “remittances” are three distinct categories of transfers. First, workers’ remittances records current transfers to nonresidents by migrants who are employed in, and considered a resident of, the countries that host them. The category employee compensation is composed of wages, salaries and other benefits earned by individuals in countries other than those in which they are residents for work performed for and paid for by residents of those countries. Finally, migrants’ transfers are contra-entries to the flow of goods and changes in financial items that arise from individuals’ change of residence from one country to another, such as movement of accumulated savings when a migrant returns permanently to the home country.