Application of MIDAS regression models in mixed-frequency data: Economic growth forecast for Turkey

Many studies up to now have been one of the important conditions that the variables subject to the analysis have the same frequency, that is, both variables are monthly or quarterly. However, because of the recent development of econometric theories, it is possible to establish a relationship between variables with different frequencies and different variants with quartiles per month. Therefore, in order to produce effective and robust estimates, new models have emerged to incorporate the same model of data with different frequencies, and these models have been included in the literature under the name Mixed Data Sampling (MIDAS). Mixed Data Sampling the most used methods in the sample are classified as Almon Polinomial Midas Regression Model, Exponential Almon Distributed Midas Regression Model, Beta Distributed Midas Regression Model, Unrestricted Midas Regression Model and Step-Weighting Method. In this study, MIDAS Regression methods were used to determine the relationships between different frequencies. The variables that were most frequently encountered in the literature for Growth Estimation were found to be the statistics such as Production Index, Non-Agricultural Employment, Inflation, Foreign Trade and these data were collected and included in the model. Almon polynomial Midas Regression Model of which it concluded that better results, and finally Turkey's 2017 economic growth forecast in this study with several MIDAS regression model attempts to economic growth forecast has predicted fourth-quarter growth.