Forecasting remittances to Mexico with a multi-state markovswitching model applied to the trend with controlled smoothness

Abstract
Remittances inflows have been associated with a reduction in the level and severity of poverty. They contribute to higher human capital accumulation, to improved access to formal financial sector services, to enhanced small business investment and to more entrepreneurship. Remittances play also an important role in contributing to the livelihoods of less prosperous people. Considering these facts, this paper proposes a statistical model to forecast remittances flows to Mexico in order to provide information for the design of policies that can help attract remittances inflows and use them productively. Here, we apply a statistical methodology based on the Multi-State Markov-Switching model with three different specifications. The model is applied to the trend of the time series data instead of the original observations with the aim of mitigating the effect of outliers and transitory blips. The filtering technique employed to estimate the trend allows us to control the amount of smoothness in the resulting trend. This method is also useful to take into account an implicit adjustment of the data at both extremes of the time series, thus providing better results than conventional filtering techniques such as the Hodrick-Prescott filter. Thus, the Markov-Switching approach captures more precisely the trend persistence of remittances and enhances both in-sample and out-of-sample forecast performance.