Asymmetric and Multi-Horizon Transmission of Fiscal Dominance and Exchange Rate Pass-Through to Inflation: Evidence from Ghana

Technical Summary: Asymmetric and Multi-Horizon Transmission of Fiscal Dominance and Exchange Rate Pass-Through to Inflation: Evidence from Ghana

Problem Statement and Motivation
This study addresses the limitations of conventional linear Autoregressive Distributed Lag (ARDL) models in characterizing inflation dynamics in Ghana (1980–2020). The authors argue that standard linear estimations impose three restrictive assumptions that obscure critical economic realities: (1) that fiscal and exchange rate effects are uniform across the entire distribution of inflation; (2) that currency depreciation and appreciation have symmetric, equal-and-opposite effects on prices; and (3) that the co-movement between fiscal/monetary variables and inflation is constant across time horizons.

The paper is motivated by conflicting empirical evidence regarding Ghana's fiscal-inflation nexus. While some studies using standard ARDL find a positive debt-inflation relationship (supporting the Fiscal Dominance hypothesis), others find a negative relationship (supporting a crowding-out mechanism). Similarly, while Exchange Rate Pass-Through (ERPT) is known to be significant, its behavior across different inflation regimes and its directional asymmetry remain under-explored in the Ghanaian context. The study posits that these contradictions arise because linear models average out heterogeneous, asymmetric, and time-varying dynamics.

Methodology
To overcome the limitations of linear ARDL, the authors employ a single empirical program integrating four distinct methodological approaches, each targeting a specific restriction of the baseline model:

1. Quantile Regression (QR): Used to test whether the coefficients of public debt servicing (PDS) and the exchange rate (EXR) vary across the conditional distribution of inflation (from the 10th to the 90th quantile). This addresses the restriction of slope homogeneity.
2. Nonlinear ARDL (NARDL): Based on the framework of Shin et al. (2014), this method decomposes the exchange rate into positive partial sums (depreciation) and negative partial sums (appreciation). It tests for asymmetric pass-through, addressing the restriction of directional symmetry.
3. Wavelet Coherence Analysis: Utilizing the Morlet wavelet basis, this method analyzes the time-frequency co-movement between inflation and its determinants. It identifies which frequency bands (time horizons) dominate the relationship at specific points in time, addressing the restriction of time-invariant frequency concentration.
4. Diagnostic Validation: The study employs the BDS nonlinearity test to validate the necessity of nonlinear methods and the Koenker slope equality test to confirm heterogeneity across quantiles. Unit root tests (DF-GLS and Phillips-Perron) confirm a mixed I(0)/I(1) integration order, validating the use of the ARDL bounds testing framework.

Key Results

• Fiscal Dominance and Crowding-Out (Quantile Results): The relationship between public debt servicing (PDS) and inflation is heterogeneous. PDS exerts a significant negative effect on inflation at lower quantiles (e.g., -1.42 at τ=0.10), consistent with a crowding-out mechanism where debt service absorbs resources that would otherwise stimulate the economy. However, this effect weakens monotonically toward the upper tail, becoming statistically insignificant at the 90th quantile (-0.28). This suggests that during high-inflation episodes, the crowding-out effect is either offset by monetization pressures or overwhelmed by other drivers.
• Exchange Rate Pass-Through (Quantile and NARDL Results):
  • Heterogeneity: The ERPT coefficient rises monotonically across quantiles, from 0.48 at the 10th quantile to 1.82 at the 90th quantile. This confirms that currency depreciations are disproportionately inflationary during high-inflation regimes.
  • Asymmetry: NARDL estimation confirms significant asymmetry. Depreciation shocks generate substantially larger inflationary effects than appreciation shocks generate deflationary effects. At the upper tail (τ=0.90), the long-run asymmetry ratio is approximately 2.5:1 (a 1% depreciation raises prices by ~2.45%, while a 1% appreciation lowers them by ~0.95%).
• Time-Frequency Dynamics (Wavelet Results):
  • Exchange Rate: Coherence between the exchange rate and inflation is highest at the 2–4 year frequency band and is amplified during crisis episodes (e.g., 1983, 2000, 2014–2017).
  • Fiscal Variables: Coherence between public debt servicing and inflation is concentrated at the 4–8 year band, indicating that fiscal transmission operates over medium-to-long structural horizons rather than immediate year-to-year cycles.
  • Money Supply: Coherence is concentrated at the short-run (1–2 year) band.

Significance and Claims
The paper claims that its multi-method approach provides a "regime-specific, direction-specific, and horizon-specific" account of inflation transmission in Ghana that linear ARDL models cannot produce.

1. Reconciliation of Literature: The study claims to reconcile conflicting prior findings on Ghana's debt-inflation relationship. It demonstrates that studies finding a positive effect likely sampled high-inflation periods where the crowding-out signal is weak, while those finding a negative effect sampled periods dominated by low-to-medium inflation where the crowding-out mechanism is strongest.
2. Methodological Justification: The authors assert that the three extensions (QR, NARDL, Wavelet) are not redundant but address conceptually distinct restrictions (distributional heterogeneity, directional asymmetry, and time-frequency concentration). The BDS and Koenker tests provide statistical validation that linear models are misspecified for this data.
3. Policy Implications:
   • Exchange Rate Management: Exchange rate stabilization is identified as the most powerful inflation control instrument, particularly during high-inflation episodes when pass-through is highest and asymmetric.
   • Fiscal-Monetary Coordination: The study suggests that fiscal and monetary policies operate on different time horizons (4–8 years for fiscal vs. 2–4 years for exchange rates), implying that coordination strategies must account for these distinct transmission lags.
   • Debt Management: The insignificance of short-term public debt (SPD) stock suggests that policy should focus on the cash-flow burden of total debt service rather than debt maturity composition.

The authors conclude that these findings offer a more nuanced understanding of inflation dynamics in Ghana and comparable Low- and Middle-Income Countries (LMICs), specifically those facing concurrent fiscal stress and exchange rate vulnerability.