The Threshold Effect of Oil Rent on Non-Oil Economic Growth in Light of Financial Development: A Case Study of Iraq

Abstract

The purpose of this study is to examine the effect of oil rent on Iraq’s non-oil economic growth and to investigate the moderating role of financial development as a threshold variable in this relationship. To measure financial development in Iraq’s economy, three key indicators—(1) the ratio of credit to the private sector, (2) the ratio of liquidity to GDP, and (3) the ratio of bank credit to GDP—were selected. These indicators were integrated into a composite index using the Principal Component Analysis (PCA) method to address collinearity and the limitations of individual measures. The index was incorporated into an extended Solow neoclassical growth model that includes financial development and oil resource variables. Time series data from 1998 to 2023 were analyzed using stationarity and cointegration tests as well as the threshold regression model. A set of diagnostic tests was also conducted to ensure the validity of the results. All estimations were performed using EViews and Stata software. The findings revealed that Iraq’s economic time series variables during the period 1998–2023 exhibited mixed stationarity characteristics—some variables, such as LnYIRAQ and LnFDIRAQ, were stationary at level, while others became stationary after first differencing. Across the four estimated threshold regression models with different financial development indicators, neither capital nor labor variables showed significant effects on non-oil growth, reflecting Iraq’s strong structural dependence on oil and institutional weaknesses. Regarding oil rent, at low levels of financial development, its effect was generally positive but statistically insignificant (except in the third model, where it was positive and significant). In contrast, at high levels of financial development, the effect of oil rent became negative and, in some cases, significant (notably in the third and fourth models). Diagnostic tests—including residual analysis, Portmanteau, and CUSUM—confirmed the stability and goodness of fit of the models, except in the fourth model, where heteroskedasticity was detected. This issue was corrected using robust standard errors, which further reinforced the negative impact of oil rent. Overall, the results indicate that even when financial indicators surpass quantitative thresholds, without improvements in institutional quality and resource allocation efficiency, adverse phenomena such as the “resource curse” and “Dutch disease” cannot be avoided.