1. DAVID UMORU - Department of Economics, Edo State University Uzairue Iyamho, Km 7 Auchi-Abuja Expressway, Iyamho Edo State, Nigeria.
2. ISEDU MUSTAFA OKHOME - Department of Banking & Finance, Ambrose Alli University, Ekpoma, Nigeria.
3. MALACHY ASHYWEL UGBAKA - Department of Economics University of Calabar, Nigeria.
4. GODWIN ELOGHOSA JEFFERY - Department of Finance and Economics, Manchester Metropolitan University.
5. BEAUTY IGBINOVIA - Department of Economics, Edo State University Uzairue Iyamho, Km 7 Auchi-Abuja Expressway, Iyamho, Edo State, Nigeria.
6. CHINYERE HELEN DEDE - Department of Public Administration, University of Calabar, Nigeria.
7. WILLIE WILFRED OKOI - Department of Economics, University of Calabar, Nigeria.
The standard linear monetary frameworks systematically miss the asymmetric changes in aggregate liquidity preferences in times of great macroeconomic stress of uncertainty. To address this structural shortcoming, the study examines the demand for money in the G7 economies (2000-2025) based on a new dual-framework: the Machine Learning-Cross-Sectionally Augmented Panel Smooth Transition Regression (ML-CS-PSTR) model and the third-order perturbed non-linear Dynamic Stochastic General Equilibrium (DSGE) model. We endogenously identify significant threshold in systemic risk using Bayesian Additive Regression Trees (BART). The macroeconomic design collapses when the composite index of economic policy uncertainty and financial crises exceeds 1.84 standard deviations. Markets below this threshold are mean reverting efficiently, trading at 28.5% per quarter. Above that, economic agents move into a high degree of precautionary saving (68.9% in the regime), inducing a regime-induced liquidity inertia and contraction for which the speed of adjustment drops to 9.2%. In addition, our non-linear DSGE variance decomposition result scientifically establishes that this slowdown is structurally attributed to the policy uncertainty shocks that persist in the economy that dominate the variance in money demand in the long run, and which are far more influential than the standard monetary policy reaction by central banks. This is supported by factor-augmented co-integration and heterogeneous causality tests that show a refined two-way causality chain between financial integration and currency volatility during threshold crises and therefore, conventional Taylor rules are not effective in this case. The threshold estimates indicate that standard Uncovered Interest Rate Parity (UIP) does not hold true during these panic periods. Based on the results, central banks should move beyond linear rate targeting to aggressive forward guidance and macro-stability policy actions and countercyclical exchange buffers for the goal of restoring market clearing mechanisms. A liquidity crisis cannot be resolved by a central bank by just increasing or decreasing the broad money supply. Policymakers should fiercely end the cloud of uncertainty if they want to restore the normal market clearing speeds. This involves coordination of fiscal and monetary policies through hyper transparent forward guidance and long-term horizons, which will help to calm markets, thereby reducing the precautionary hoarding elasticity back to baseline levels.
Money Demand; Non-Linear Panel Econometrics; ML-CS-PSTR; Economic Policy Uncertainty; Dynamic Stochastic General Equilibrium (DSGE); Cross-Sectional Dependence; G7 Macroeconomics.