Decoding Fiscal Dominance: A Systematic Review of Machine Learning-Informed Evidence on Inflation Persistence in Pakistan

Abstract

This study presents a PRISMA-based Systematic Literature Review (SLR) examining the relationship between fiscal dominance and inflation persistence in Pakistan through the lens of machine-learning-informed macroeconomic research. The review synthesizes evidence from peer-reviewed articles, central bank working papers, and international financial institution reports published between January 2015 and January 2026. Literature was identified through systematic searches of Google Scholar, Scopus, Web of Science, IMF eLibrary, World Bank repositories, and State Bank of Pakistan publications using predefined Boolean search strings. Following screening and eligibility assessment procedures, 25 studies were included in the final synthesis. The reviewed evidence consistently identifies fiscal deficit monetization, broad money growth, exchange-rate depreciation, and energy-price shocks as the dominant drivers of inflation persistence in Pakistan and comparable developing economies. Findings from Random Forest, LASSO, and Long Short-Term Memory (LSTM) studies suggest that fiscal variables frequently exhibit greater explanatory power than conventional monetary indicators during periods of fiscal dominance. The review further highlights significant provincial and income-group heterogeneity in inflation outcomes, with lower-income households experiencing disproportionate welfare losses. The study concludes that sustainable disinflation requires stronger fiscal-monetary coordination, enhanced central bank independence, improved revenue mobilization, and targeted social protection measures. The findings contribute to the growing literature on machine-learning applications in macroeconomic policy analysis and provide a structured evidence base for inflation-management reforms in Pakistan.