An Econometric Framework for Evaluating Safety on Tolled Motorways

This thesis expands upon econometric and traffic flow theories, to evaluate the economic performance of tolled motorways, by exploring case studies in Sydney, Australia, through the integration of complex datasets that capture consumer behaviour, traffic dynamics and crash events. The objective is to refine the calculation of societal impacts borne by road users during disruptions, through the incremental integration of data of varying spatiotemporal resolution and the estimation of treatment effects, using causal inferential modelling. At a “Network Level”, an eight-year longitudinal analysis is undertaken, to assess temporal changes in casualty costs relative to toll revenue. This ratio serves as a novel proxy for crash exposure, offering insights into the evolving safety-economic trade-offs, during periods of significant disruption, e.g. every $1 of revenue incurred $0.09 in casualty costs, during April 2020. A difference-in-differences (DiD) analysis was undertaken to model the temporal shift of these trade-offs. The “Segment Level” case study utilises a derived database, combining 15-minute gantry volumes with vehicle telematics to quantify the societal costs of planned versus unplanned disruptions. Applying principles of network resilience; this analysis benchmarks baseline performance against disruption-induced deviations, to calculate the proportional change in value of travel time imposed on road users, relative to their seasonal trip utility. The “Nodal Level” case study introduces 30-second loop detector data to evaluate lane-by-lane speed profiles during disruptions. Systematic DiD scenario testing was undertaken to model traffic dynamics between heterogenous disruptions (i.e. road works vs. road works with accident) to enhance the detection of shocks, before, during and after disruptions. Collectively, this thesis underscores how systematically increasing data resolution refines the economic evaluation of tolled motorways during disruptions. Key contributions include: (1) linking macro- and microeconomic metrics to inform motorway performance, (2) the empirical benchmarking of utility and the internalised costs absorbed by road users, and (3) retrospective evaluation of traffic response, using dynamic safety-economic performance metrics. Through this incremental approach, quantitative insights are offered into the relationship between the amount of rigour and data required, to adequately inform categories of policy. For example, when is a macro analysis adequate or what is the relative error between datasets of differing spatiotemporal resolution when investigating the same disruption? The findings advocate for the integration of near real-time datasets in toll road diagnostics, to ensure road users receive value for money, particularly when disruptions alter expected utility. This research provides a globalised framework for policymakers to optimise interventions and minimise societal costs, for tolled motorways.