Abstract

The financial crisis of 2007/2008 sparked a wave of criticism of economic theory. These assaults are based on four basic criticisms: economists failed to predict the worst financial catastrophe since the Great Depression; authorities allowed bubbles to build unchecked; poor banking supervision; and macroeconomic policy models that were out of touch with reality. The DSGE models (dynamic, stochastic, and general equilibrium) and its two simplifying hypotheses: the representative agent and rationality, are the target in the case of macroeconomic models. Models of Dynamic Stochastic General Equilibrium (DSGE) are frequently employed as a tool for formulating policy. When these models failed to foresee the 2008 financial crisis and failed to address subsequent policy issues, they lost their prominence. Meanwhile, a DSGE model counterpart known as Agent-Based Modelling (ABM) emerged. In this study, an attempt is made to adapt a macroeconomic agent-based model of the Pakistani economy. For this, we employed an economic model for simulation and a surrogate modelling approach to investigate the parameter space of the macro agent-based model. Utilizing data from Pakistan, we fixed three model parameters before letting the remaining parameters undergo random exploration. Through an agent-based model with credit and capital, we attempted evaluate the monetary policies implemented by the monetary authority while taking into account the interbank market, the monetary authority combined with machine learning calibration techniques. We performed two monetary policy experiments in a calibrated agent-based model and investi- gated the impact on GDP, unemployment, and inflation. Our model suggests that agent-based macroeconomic models might be used to analyze monetary policy in Pakistan but there are computational limitations to addressed for better and improved simulation results. The comparison of the effects of monetary policy on different macroeconomic variables can be done using the same paradigm.