A Comparison Between the ARDL and the LSTM for Exchange Rates

  • สุพัตรา ศิริลาภ
  • สมพร ปั่นโภชา
Keywords: ARDL, Deep Learning, Recurrent Neural Network, LSTM


This study aims to study the efficiency of forecasting the exchange rate of Thai baht against the US dollar using the model based on the concept of purchasing power parity. The monthly data from January 2010 to December 2021, 144 months are used for estimation. Two estimation techniques are employed in the study. The first technique on is Auto Regressive Distribution Lag Model (ARDL) approach with Bound test statistic. The other one is Long Short-Term Memory Model (LSTM)  which is non-linear model.

From ARDL results, we can conclude that an increase relative output lead of baht appreciation 0.714%. In terms of the error estimation comparison between ARDL and LSTM ,the result shows that ARDL model has perform better than LSTM.


Most read articles by the same author(s)

1 2 3 4 5 > >>