Running urban rail services can be costly; in some contexts, it requires subsidy. Therefore, ensuring that costs are kept in line is vital as it affects government expenditure and passenger fares. This condition motivates studies that understand the cost structure and whether firms operate efficiently. The empirical work presented in this thesis centres on urban rail in Japan as its primary focus. This thesis comprises three interrelated research studies. Research Study 1 aims to understand the cost structure of each urban rail mode in Japan. Research Study 2 explores the ownership effect on cost efficiency. Research Study 3 further explores ownership and other effects on cost efficiency, service effectiveness, and cost effectiveness. This thesis utilised the trans-log cost function and DEA-Tobit regression to achieve the research aims. The trans-log cost function is parametric, while DEA-Tobit regression is semi-parametric. Nevertheless, they are two widely used methods for deriving performance, especially efficiency. There are lessons from this thesis, especially on the cost structure, mode differences, and ownership effects. First, traffic density and scale affect different performance dimensions (i.e., cost efficiency, service effectiveness and cost effectiveness) in different ways. Second, mode affects different performance dimensions in different ways. Third, Returns to Density (RTD) and Returns to Scale (RTS) vary between over-ground, monorail, and under-ground. Fourth, private firms are profit-maximising entities but not necessarily cost-efficiency maximisers. Fifth, measuring all the performance dimensions and interpreting the results relative to each other is essential. These findings are essential for firms, regulators, and funders. Given the interest in the empirical performance of private urban rail firms, we suggest future research investigate how they perform in cost efficiency, service effectiveness and cost effectiveness in other regions. We also hope that future empirical research will clarify the RTD and RTS of urban rail modes.
