Does Pavement Degradation Follow a Random Walk with Drift? Evidence from Variance Ratio Tests for Pavement Roughness

Abstract

Planning agencies increasingly use pavement management systems (PMS) to determine an optimal preservation strategy for their roadway assets. To develop cost-effective resource allocation policies, it is important that a PMS embed pavement degradation models that accurately depict its progression over time. Presently, PMS frameworks use two broad classes of methods (Markov chains and simplified, trend-stationary regression models) to project pavement degradation. These approaches make contradictory assumptions regarding (1) the degree to which variation is aleatory/epistemic, and (2) the long-term persistence of sudden changes in pavement distress. Consequently, this research constructs a panel data variance ratio test to evaluate if pavement degradation conforms to a hypothesis that convolves the assumptions of the two prevailing approaches: a random walk with drift that captures relevant exogenous information. The authors apply their model to publicly available data on pavement roughness, one pavement distress mechanism of primary interest to planners, as part of the Federal Highway Administration’s Long-Term Pavement Performance (LTPP) program. The case study results are unable to reject the null hypothesis that pavement roughness follows a random walk with drift, a model structure that contradicts the current assumptions of PMS platforms. The methods developed by the authors offer decision makers an opportunity to augment their current PMS approaches, because the misspecification of pavement degradation will cause such decision-support tools to select suboptimal allocation policies.