CONCLUSIONES Y RECOMENDACIONES

As widely recognised in the literature, several departures from ideal first best conditions exist in reality. In the case of public transport pricing, the most evident and analysed case is that buses or trains compete with underpriced cars, as cultural, technical, political or social constraints impede the setting of marginal cost road pricing. When this inefficiency is present, the optimal pricing analysis in public transport is referred to as second best pricing (although as previously mentioned, the second best concept may encompass several other distortions both within and outside the transport sector). The classical argument is that if cars are underpriced, there is an excess of car travel, therefore it would be welfare improving to reduce the public transport fare in order to attract some drivers to use trains or buses, in turn reducing the level of congestion and other traffic externalities on the road network. This is a second economic-based rationale to subsidise public transport, after the economies of scale (first best) argument7

7 _{Other arguments in favour of subsidising public transport include pursuing distributional or social}

objectives and option values, which are not treated in this thesis (see Kerin, 1992; Preston, 2008).

(Preston, 2008; Parry and Small, 2009). Therefore, as argued by Small (2008), a conclusion from first best and second best fare analyses is that congestion charging could be seen as a way to reduce

27 the financial needs of public transport, since an optimal road charge should decrease the subsidy required for public transport, even if the revenue from road pricing is not hypothecated to public transport.

Formal proofs that an alternative mode should be priced below marginal cost when cars are priced at average instead of marginal cost can be traced to Lévy-Lambert (1968), Marchand (1968) and Sherman (1971). The idea, linked to competitive neutrality, was extended by Glaister (1974), who finds a second best bus fare below marginal cost, not only in the peak but also in the (congestion free) off-peak period, the latter due to two effects - a low off-peak bus fare can attract peak car users, and peak bus users are attracted to travelling by bus during the off-peak, which relieves pressure in the peak, and therefore decreases the peak bus fare, which in turn attracts more car travellers into public transport.

A slightly different approach is introduced by Jackson (1975), who instead of calculating the optimal second best fare, estimates the optimal second best subsidy directly, assuming the underpricing of highway travel, and that average cost per bus user is constant. The optimal fare subsidy depends on the level of congestion on the road, the own cost elasticity of transit demand, and the cross cost elasticity of demand between private and public transport. Interestingly, Jackson (1975) proposes a method to determine the optimal subsidy for bus speed improvements, instead of covering fare reductions, with the result depending on how large the increase in operator cost is to achieve improvements in speed. Illustrative examples show that the welfare improvements largely rely on, and increase with, the degree of congestion associated with highway travel, and the cross elasticity of demand with respect to the generalised cost of public transport. Jackson’s case for second best welfare maximisation through an increase in the quality of bus service, as opposed to a fare reduction, parallels the contributions of Mohring (1972) and Turvey and Mohring (1975), who identify a first best

28 justification for bus subsidies due to the reduction in bus waiting times if frequency is optimally adjusted as demand grows, which is another way of speeding up bus travel.

In terms of congestion externalities, most transport pricing studies focus on automobiles (and trucks in the freight context) as the major source of traffic congestion imposed on all modes that share the right of way; however this is not necessarily the case for high frequency bus services that may slow down both cars and buses. The effect of this congestion effect of buses on cars was analysed by Else (1985), who shows the impact of the external congestion cost of public transport over the second best fare and subsidy. Bus congestion by itself increases the optimal fare and reduces subsidy, however, using British data, Else (1985) shows that even when recognising that public transport contributes to congestion, the optimal fare does not cover operating cost, and an optimal subsidy is required.

Finally, we mention the work of Parry and Small (2009), who show that substantial gains in social welfare are accrued from diverting car drivers into public transport (second best argument) in peak periods, whereas the case to subsidise fares due to the reduction of users costs (scale economies – first best argument) is stronger in the off-peak. When there are two public transport modes (bus and rail), they consider that reducing the fare on one has consequences over the other; for example a drop in the rail fare would attract bus passengers, resulting in increased waiting and access times scale economies, negative effect), but decreased bus operator cost, in-vehicle crowding and externalities (positive effect).

In summary, the above suggests that setting public transport fares below average operator cost is supported by most of the formal analysis of pricing, resulting in the call for an ‘optimal’ subsidy regardless of whether it is based on first best or second best grounds. Despite the rigorous analytical approaches and empirical evidence, the extant literature has a number of limitations associated in particular with the omission of non-

29 motorised modes such as walking and cycling, and the distortionary effect of bus subsidies, as identified by Kerin (1992), who more precisely states that “the results of the second-best pricing studies are derived under conditions that are probably unduly favourable to second-best bus pricing. If the key omitted factors could be incorporated into the trade-off process, the optimal second best subsidy level would probably be much lower than that suggested by existing formal models” (Kerin, 1992, p.39). Some of these factors have been accounted for in more recent research, such that possible inefficiencies associated with subsidy (Section 2.2.3), the existence of tax distortions and their interaction with the transport system (Section 2.3.4), and the impact of bus congestion on travel times and operation costs (Chapter 5). The influence of non-motorised transport on optimal pricing decisions is addressed with a multimodal pricing model for the maximisation of social welfare in Chapter 3.