Determinants and stability of demand for money: The case of Turkey

Uncertainty has become a phenomenon that directs the expectations of economic agents about the future and affects the decision-making mechanisms regarding money demand as a result of recent global and national developments. The effective forecasting of money demand under conditions of uncertainty is of decisive importance in minimizing the instability effects of possible shocks and maintaining economic equilibrium. Friedman's Monetary Volatility Hypothesis stands out as one of the fundamental theoretical frameworks explaining this dynamic, suggesting that economic agents facing uncertainty tend to hold more liquidity. The aim of this study is to examine the behavior of money demand in Turkey under uncertainty conditions in the context of Friedman's Monetary Volatility Hypothesis through symmetric and asymmetric cointegration and causality analyses. For this purpose, variable-based uncertainty series are constructed using ARCH and GARCH methods and index-based uncertainty series are constructed using the Atta-Mensah approach. The study covers two different periods: the main period January 2001 - August 2024 and the sub-period January 2008 - December 2020. Under the assumption that money demand is equal to money supply, cointegration relationships between real M1 and real M2 monetary aggregates representing money demand and uncertainty series are analyzed symmetrically using ARDL and asymmetrically using NARDL methods. The symmetric causality relationship between the uncertainty series and the real monetary aggregates used to represent money demand is analyzed using Toda-Yamamoto causality analysis under the Seemingly Unrelated Regression (SUR) method and the asymmetric causality relationship is analyzed using Hatemi-J asymmetric causality analysis. The findings obtained from ARDL and NARDL methods reveal that the effect of uncertainty on money demand in the long and short run is more intense and generally variable in the context of asymmetric tests. The process by which economic agents rationalize their money demand decisions under uncertainty conditions differs depending on the type of uncertainty. Moreover, according to the results of both causality analyses, the real M1 monetary aggregate is more sensitive to uncertainty conditions compared to real M2, and this sensitivity becomes more pronounced for positive shocks to real M1. As a result of the analyses, there is strong evidence for the validity of Friedman's Monetary Volatility Hypothesis for Turkey, especially in the context of causality tests.