In a panel of West African countries, we investigate whether data on immigrant remittance flows can be used to improve on predictive accuracy of aggregate demand in a systematic way. The results of the prediction experiments are compared to traditional significance tests of asymmetric error correction and of the exogenous remittance variable. We find that there is a considerable discrepancy between statistical hypothesis testing and the results from the forecast comparison. In particular, while remittances yield significant coefficients for at least some of the accounts aggregates, they do not contribute to improved forecasting accuracy.
remittances time series prediction cointegration net exports