Fairness and toll roads

Date published: 2010-Apr-05

Sarah Truen

The Gauteng Freeway Improvement Project tolls are one of the first big tests of the policy of "user pays" for infrastructure which has been introduced by government. In theory, making the user pay for the infrastructure that they benefit from has the advantage of fairness, and helps to ease funding constraints that could otherwise prevent the construction of economically important infrastructure. However, in practice, both the fairness and economic sense of a given user pays project depends on the details of how it is implemented - and this is where problems begin to arise with the GFIP.

Sanral hasn't put much information into the public arena, so we can only get initial indications of the size of the problem. However, looking at build costs and the proposed capital repayment schedule alone, it seems pretty clear that something odd is going on.

The GFIP has cost Sanral R20 billion, which has allowed it to upgrade 185 kilometres of highway, and increase its capacity by (on average), one lane in each direction. This works out at a per kilometre road cost of about US$15.4m. Comparing the cost of the GFIP to other projects which are reasonably similar yields some very interesting results. First, we compared the GFIP to ten highway projects undertaken by the World Bank, mainly in developing countries. The World Bank projects achieved an average per kilometre cost of US$4.7m per kilometre, or 69% cheaper than the GFIP. Then we compared them to build costs in the United States. On a sample of ten US highways, the average per kilometre build cost was US$7.5m - 49% cheaper than the GFIP. Its pretty difficult to compare the costs of different highway projects, as much of the cost depends on factors such as how difficult the terrain is, how many interchanges there are, and how expensive the land involved is. However, the size of the cost discrepancy between the GFIP and (reasonably) similar international projects is so big as to raise serious questions about the build costs of the GFIP.

From this analysis, it looks like there might be a problem with how much the user is being asked to pay for - if costs are grossly inflated, then the fairness of making the user pay is a lot harder to justify. However, even if we accept that the R20 billion cost of the GFIP is fair, a quick calculation suggests that at 10% interest, Sanral only needs to raise R193 million a month to repay the loan over the planned 20 years. Sanral's projected revenue from the GFIP tolls is R300 million per month. What exactly is the remaining R100 million or so going towards? This starts to look less like an incidence of user pays, and more like an incidence of user cross-subsidises.

If these are the concerns that arise if one looks simply at the data that Sanral has put into the public arena, it would be fascinating to see what full public disclosure would reveal. At present we have little or no idea as to how toll rates are being determined, for example. And all these questions arise before one deals with the issues of the impact of the toll scheme on South Africa's largest provincial economy.