How could data and AI improve three methods for assessing the impact of new transport projects

Data and AI could improve the methods to assess the impacts of new transport

Transport infrastructure is a powerful driver of economic and social change. Governments and planners need reliable methods to assess the likely impacts of major investments in roads, rail, and public transport. Several analytical tools exist for this purpose, each with their own strengths, limitations, and levels of complexity.

This article focuses on three key methods: Gross Value Added (GVA) analysis, Computable General Equilibrium (CGE) modelling, and Cost-Benefit Analysis (CBA), while also exploring emerging opportunities to integrate new data sources and artificial intelligence (AI).

Traditional GVA-based approaches

Gross Value Added (GVA) analysis estimates the economic output generated by a transport investment. Typically, this is calculated by assessing the expected increases in employment, productivity, and activity in affected sectors or regions. GVA estimates are often derived from observed correlations between accessibility improvements and economic indicators.

This method is particularly popular because it produces concrete numbers that can be directly tied to GDP and employment. It is intuitive and policy-relevant, especially for politicians and decision-makers seeking tangible economic returns.

However, GVA-based assessments face key limitations:

• Causality vs correlation: They often rely on before-and-after comparisons or modelling based on assumed elasticities, which do not always prove a direct causal link.

• Double counting: There is a risk of overlap with other measures like travel time savings, leading to inflated estimates.

• Exclusion of distributional impacts: These methods rarely distinguish between who gains and who loses across sectors and regions.

A data-driven, AI-enhanced GVA method—GVA 2.0—addresses these limitations by leveraging real-time, granular data sources. AI tools can analyse anonymised mobile phone data to track actual commuter flows and job shifts following transport improvements. Urban mobility dashboards powered by machine learning can continuously monitor changes in business activity and local economic output. These dynamic insights allow planners to attribute economic gains more accurately to infrastructure investments and detect unintended effects earlier.

By moving beyond static comparisons, AI-augmented GVA approaches offer more defensible, up-to-date economic narratives. For example, they can identify the formation of new economic clusters or monitor the displacement of economic activity in surrounding areas—critical for robust, distribution-aware planning.

GVA methods remain useful in providing headline indicators but now benefit greatly from data-driven enhancements that increase transparency, attribution quality, and policy relevance.

Cost-benefit analysis (CBA)

Cost-benefit analysis (CBA) is a long-standing and widely used method that compares the total expected costs and benefits of a project over its lifetime. It aims to determine whether the benefits to society exceed the costs of construction, maintenance, and operation. When applied rigorously, it offers a structured and transparent framework for project evaluation.

Key elements of transport CBA include:

• User benefits: Primarily travel time savings, vehicle operating cost reductions, and improved safety.

• Reliability and environmental impacts: Increasingly, CBAs account for travel-time reliability and emissions reductions.

• Monetary valuation: Non-market impacts, such as noise and greenhouse gas emissions, are monetised using shadow pricing or revealed preferences.

Despite its widespread use, CBA has faced criticism, particularly in the context of large transport projects. Key challenges include:

1. Mismatch with political priorities: CBA typically evaluates national welfare but may not address political concerns about jobs, local development, or regional equity.

2. Scope limitations: Standard CBA often omits wider economic impacts such as agglomeration or land use changes due to modelling constraints.

3. Static assumptions: It rarely captures dynamic changes in behaviour, such as the relocation of households or businesses, over time.

Enter CBA 2.0: AI-enhanced cost-benefit analysis offers the potential to expand both the depth and real-world applicability of the method. AI tools can integrate real-time sensor data, mobile GPS, and weather information to dynamically model traffic flows and travel time savings. In Singapore, for example, AI is used to adjust public transport schedules based on real-time demand, providing a practical feedback loop for cost-benefit inputs.

AI-powered predictive analytics help simulate behavioural changes, such as the likelihood of switching modes of transport or adjusting commuting patterns in response to new infrastructure.

These insights help generate richer and more accurate estimates of long-term impacts, especially under different future scenarios. Advanced CBA models can also test various environmental and economic conditions using AI scenario tools, offering planners probabilistic and stress-tested views of possible outcomes.

By embedding these tools into standard practice, CBA becomes more adaptive, reflective of local nuances, and aligned with modern policy challenges such as climate mitigation and urban equity. CBA 2.0 allows decision-makers to understand not just what is likely to happen, but what could happen under a range of intelligent, data-informed conditions.

Computable general equilibrium (CGE) modelling

CGE models simulate the entire economy’s response to a transport investment by accounting for interlinkages between sectors, households, and government. They estimate how changes in travel costs influence wages, prices, employment, and production across regions and industries.

The advantages of CGE models include:

• Capturing indirect and long-term effects: They can show how transport affects land use, productivity, and sectoral growth.

• Consistent with economic theory: CGE models use a system of equations grounded in microeconomic behaviour and market interactions.

• Scenario flexibility: They are well suited for testing the economic effects of different investment options and policy changes.

However, CGE modelling is data-intensive, complex, and difficult to validate. Key criticisms include:

• Sensitivity to assumptions: Results depend heavily on the chosen elasticities and input-output tables.

• Transparency and replicability: The models are often seen as ‘black boxes’ to non-specialists.

• Implementation barriers: High expertise and computing resources are required.

CGE 2.0 integrates AI and dynamic data sources to improve model accuracy, transparency, and usability. AI tools can automate the calibration process and test large numbers of input assumptions quickly, revealing sensitivity in outcomes. Moreover, AI helps to ingest and process vast mobility datasets from GPS systems, smartcards, and IoT sensors, providing richer representations of household and firm behaviour.

Cities like Dubai are already applying AI-powered CGE approaches in their smart mobility plans. These tools simulate infrastructure scenarios under varying economic and environmental assumptions, helping policymakers anticipate knock-on effects like land use changes, regional wage shifts, or sector-specific growth. Machine learning algorithms can also detect anomalies or inconsistencies in CGE simulations, increasing confidence in the models.

Crucially, the use of AI does not replace economic theory but complements it by making the models more empirically grounded and scenario-rich. This allows CGE analysis to serve as a strategic foresight tool that supports more resilient, long-term planning.

Conclusion

No single method perfectly captures all the impacts of a transport investment. GVA methods are accessible but limited in depth. CBA remains the cornerstone of project appraisal, offering transparency and rigour, but it must be applied carefully and supplemented where needed. CGE models provide a broader view of economic changes but come with high complexity.

However, version 2.0 of each method—augmented by AI and new data sources—offers substantial improvements. These next-generation approaches enable better attribution, dynamic responsiveness, and richer scenario planning. As smart cities deploy more advanced transport systems, planners can leverage AI for real-time monitoring, predictive modelling, and sustainable design.

Ultimately, effective planning will depend on a balanced approach that blends technical robustness with policy relevance. Policymakers should aim for a layered toolkit—using GVA, CBA, CGE, and AI-enhanced methods in combination—to make well-rounded and forward-looking infrastructure decisions.