Dynamic Panel Data Analysis of Road Capacity and Traffic Demand: Evidence from the Rawalpindi Division, Pakistan

This study estimates traffic demand response to road capacity changes in Rawalpindi Division, Pakistan, using panel data on Vehicle-Kilometers Traveled (VKT) and lane-kilometers. It provides subnational evidence from a South Asian context where link-level panel data are rarely available.

A dynamic panel dataset comprising 69 road segments observed over nine years (2014–2023) was analyzed. Arellano–Bond and Blundell–Bond System GMM estimators were applied to address potential endogeneity, unobserved link heterogeneity, and the dynamic adjustment in traffic demand. All continuous variables (VKT, lane-kilometers, fuel price, vehicle ownership, and population growth rate) were log-transformed to estimate elasticities and stabilize variance.

The preferred System GMM yields an elasticity of 0.967 for lane-kilometers, implying that a 1% increase in lane-kilometers is associated with an approximately 1% increase in VKT. Fuel price, vehicle ownership, and population growth rate are statistically insignificant in the preferred model. Diagnostic tests support instrument validity and indicate no evidence of second-order serial correlation in differenced residuals.

The near-unit elasticity is consistent with induced-demand mechanisms, suggesting that capacity expansions are associated with higher traffic volumes and erode congestion relief. The weak effects of fuel prices and vehicle ownership reflect Pakistan’s institutional and behavioral context, including regulated fuel prices and a vehicle fleet dominated by motorcycles, along with limited modal alternatives.

Road capacity expansion in Rawalpindi Division is strongly associated with higher traffic volumes. Supply-only strategies are therefore likely to deliver limited long-run congestion relief unless accompanied by multimodal mobility measures and demand-management policies.