Government Capital Expenditure and Private Non-Oil Gross Fixed Capital Formation in Saudi Arabia: Evidence from an ARDL Approach

Abstract

Purpose

This study examines the dynamic effects of Government Capital Expenditure (GCE) on Private Sector Non-Oil Gross Fixed Capital Formation (GFCF) in Saudi Arabia, with particular attention to the pre-Vision 2030 period, the 2015–2016 oil-price shock, Vision 2030, and the post-COVID recovery.

Methodology

Using monthly data from March 2011 to September 2025, the study employs the Autoregressive Distributed Lag (ARDL) bound testing approach to estimate long- and short-run relationships among the variables. Further, pairwise Granger causality and VAR Block Exogeneity Wald tests assess short-run predictive relationships, while Chow breakpoint tests examine structural stability around major events.

Findings

The ARDL bounds test confirms cointegration, as the OLS F-statistic of 7.597 exceeds the I(1) bound at the 1% level. HAC-based evidence is less pronounced (F = 3.794), while the HAC error-correction t-statistic (−3.573) narrowly exceeds the 1% critical bound (−3.43). GCE has a significant negative long-run effect on private non-oil GFCF (−1.943; HAC SE = 0.276; p < 0.001; 95% CI [−2.492, −1.394]), whereas trade openness (3.340; p = 0.005) and private-sector credit (1.629; p = 0.036) have positive effects; Brent prices are insignificant. The error-correction coefficient (−0.357) indicates a 35.7% monthly adjustment towards equilibrium. No significant short-run predictive causality exists between GCE and private GFCF. The Chow test finds no break in April 2016 but identifies a significant structural break around March 2020 (p < 0.001).

Conclusion & Practical Implications

The findings suggest that government capital expenditure crowded out, rather than crowded in, private non-oil investment during the study period, with important implications for public-investment policy under Vision 2030. The study contributes updated evidence through September 2025 by integrating ARDL estimation, extensive diagnostics, causality testing, and structural-break analysis.