The causality relationship between remittances and the real effective exchange rate: the case of the Kyrgyz Republic

Abstract

We assessed the causality relationship between the inflow of remittances and the real effective exchange rate (REER) of the Kyrgyz Republic, one of the most remittance-dependent economies in the world. We utilised the procedure suggested by Hong [15]. In the first step, we estimated univariate generalised autoregressive conditionally heteroskedasticity (GARCH) models for the logarithmic difference of the variables and saved the centred standardised residuals and their squared values. In the second step, we computed the sample cross-correlation function (CCF) between the standardised residuals and squared standardised residuals. Finally, we calculated Hong’s [15] Q-statistic and compared it to the upper tailed critical value of N(0;1) at an appropriate level. Descriptive statistics showed more volatile standard deviations for remittances compared with the REER. The skewness values indicated that increases are more likely to occur for the REER and decreases are more likely to occur for remittances. High kurtosis values suggested the existence of heavy tails in the return distribution. The skewness and kurtosis values showed that the returns are not normally distributed. At the 1% significance level, the Lagrange Multiplier (LM) test for autoregressive conditional heteroskedasticity (ARCH) rejected the null hypothesis of ‘no ARCH effects’ for both variables. At the 1–5% significance level, the augmented Dickey–Fuller (ADF) test rejected the null hypothesis that a unit root is present in the return series of both variables. The test for a structural break (a supremum Wald test) in the return series, without imposing a known break date, rejected the null hypothesis of no structural break at the 5% and 1% levels for the REER and remittances, respectively. The estimated break date for the REER was May 2015, and for the remittances, it was February 2008. The logarithmic differences of seasonally adjusted real monthly data for the period extending from January 2005 to December 2017 were used in the estimations. While the existing literature has focused mostly on the impact of remittances on the exchange rate (a unidirectional causality in the mean from remittances on the exchange rate), we examined how remittances cause the REER in the mean and variance as well as how the REER causes remittances in the mean and variance. The derived results did not reveal evidence of causality in the mean or variance from the remittances to the REER for the case of the Kyrgyz Republic. That means the changes and volatility of the remittances did not contain useful information for predicting the REER. On the contrary, the REER caused remittances in the mean and variance. That means past information and the volatility of the REER were useful for predicting remittances’ returns and volatility.