Trend Inflation and the Costs of Price Dispersion in a Fiscal DSGE Model

Working paper 923

Abstract

Most inflation-targeting frameworks implicitly endorse a positive trend inflation rate as desirable. Yet, consensus on an optimal inflation target remains elusive, particularly for economies such as South Africa, which employs a broad target band of 3% to 6%. Using a medium-scale New Keynesian DSGE model with fiscal dynamics, this paper demonstrates that even moderate levels of trend inflation (between 4% and 6%) entail substantial real resource costs due to increased price dispersion. Elevated trend inflation amplifies resource misallocation between supply and demand sectors, flattens the Phillips curve, increases markup variability among firms, and weakens the economy’s responsiveness to shocks. Consequently, steady-state output declines, welfare losses escalate nonlinearly, and the sacrifice ratio—measured as the output cost of reducing inflation—rises markedly. Bayesian estimation on South African data (2000Q1–2024Q2) confirms that excluding trend inflation in a Taylor rule framework exaggerates policy inertia and understates the aggressiveness required to stabilise inflation. Allowing trend inflation to drift toward the upper bound of the target range thus increases macroeconomic volatility and erodes welfare. In contrast, anchoring inflation expectations closer to 3% mitigates price dispersion, restores a steeper Phillips curve, and improves macroeconomic welfare outcomes.