PREDICTION OF FOREIGN DIRECT INVESTMENT: AN APPLICATION OF LONG SHORT-TERM MEMORY

Abstract

In any country, Foreign Direct Investment(FDI)plays a crucial role in the development process through the transfer of financial resources, innovative management techniques, technology and raising productivity. Developing countries like Indianeeds inflow of foreign capital to fasten economic growth and development. It also helps the policy makers to fine tune their economic policies to attract more FDI and take comparative advantages in the competitive world. However, today’s globalized world makes it more complex, volatile and noisy in nature. Therefore, in reality, it is very difficult to predict the future values of FDI. Linear models sometimes can’t capture the complex patterns present in the data. It is expected that estimation of FDI through Deep learning method would not only be able to capture such volatility more efficiently but also be able to predict future inflow of FDI more accurately than conventional forecasting techniques. In this paper, we have implemented Long Short Term Memory (LSTM) algorithm to predict the future inflow of FDI. It has been shown that LSTM proves it’s supremacy over Autoregressive Integrated Moving Average(ARIMA), Generalized Autoregressive Conditional Heteroscedasticity(GARCH) and other linear forecasting models.