Many studies in the literature assessing port choice have relied on disaggregated data to produce their estimates, especially those using discrete choice models as their methodology. Although this approach has been useful for understanding the relevance of diverse attributes on port choices, models with disaggregated data have limitations in forecasting port market shares expressed in tons of cargo handled by ports. To address this issue, we estimate port choice models for imports and exports, using aggregated open data. We use Colombia as case study, using the entire record of import and export transactions of several products in one year, in the country’s four main public ports. This approach allows us to forecast the number of tons handled by each port and type of product. The models are then used to evaluate the impact of a project to improve the navigability of the Magdalena River, the country’s main waterway. Results suggest that freight rates are the most relevant factor in port choice. Our findings show how increasing freight rates and transit times decrease the likelihood of selecting a specific port zone, whereas increasing the call frequency of shipping lines at a port zone increases its use. Results also suggest that the willingness to pay for shorter transit times is higher for imports than for exports, while the willingness to pay for higher call frequency of shipping lines is similar for both. According to elasticity analysis, the call frequency of shipping lines is less sensitive to port choice than transit time and freight rates. Finally, we evidence that using an aggregate specification to forecast the tons by type of cargo is a powerful tool for port authorities and policymakers.
