What are relations between the domestic macroeconomic variables and the convertible exchange rates?

Abstract

Worldwide foreign trade operations and financial investment activities are realized as convertible currencies. The level of exchange rate volatility may be higher in developing countries than in developed countries. This situation can be seen both in the real sector and in the financial sector. In the real sector, especially companies that meet their raw material and intermediary needs through import are affected by the exchange rate volatility, while the financial sector is more affected by exchange rate volatility in the markets with weak-efficiency levels. These determinant attributes reflect the characteristics of developing countries. Therefore, the identification of parameters that can provide knowledge about changes in exchange rates is important for both the real sector and the financial sector. Also, this knowledge is also important regarding increasing the effectiveness of exchange rate interventions, one of the instruments of monetary policy in modern central banking practices. This chapter aims to attain explanation capacity of domestic macroeconomic factors of convertible exchange rates and rules for the application made by OneR algorithm which is one of the data mining methods and the machine learning techniques. The reason for using only domestic macroeconomic variables and the exclusion of international macroeconomic variables in the study is that it is more frequent to attain knowledge about domestic macroeconomic variables which are estimated within the countries. Thus, it is aimed to increase the frequency of observing convertible exchange rates with the rules acquired by the OneR algorithm. It is also aimed to investigate whether the exchange rate movements included in this study can be modeled by using only domestic macroeconomic variables as a glocal approach. EUR/USD, GBP/USD, JPY/USD, and TRY/USD exchange rates are analyzed within the scope of the chapter. The findings of the chapter show that (1) the problem of estimation of the exchange rate movements is insufficient to solve by OneR algorithm; (2) it is seen that the success rate of the models with a relatively small number of input variables is higher in this application; therefore, the importance of the use of lean models is supported by the results of the chapter; and (3) in terms of the primary aim of the survey, Turkey’s domestic macroeconomic variables are not sufficient to explain convertible exchange rates. As the reasons for these findings, it can say that the Turkish economy is a developing economy and that the economy is small compared to developed country economies.