Taxpayer funding for toll roads

The public outcry over the proposed tolls for the Gauteng Freeway Improvement Project (GFIP) have substantially delayed Sanral’s toll implementation plan. They also pose a more fundamental question: if the user-pays principle is no longer going to be used to finance road infrastructure, then who is in fact going to pay? In this year’s budget, we got our first hint at an answer to this question, and not unexpectedly, its the taxpayer.

For the coming financial year, National Treasury has allocated an additional R5.8 billion to Sanral, specifically to defray the cost of the GFIP. This is over and above the R5.3 billion already allocated to Sanral for capital expenditures on other parts of the road network, but represents only around 29% of the total GFIP project debt of R20 billion. No future transfers to Sanral for the GFIP are currently budgeted for, but if the tolling plan falls apart or consumers refuse to pay, we can expect more transfers in future financial years.

What exactly is the impact of this going to be? If the GFIP is funded by budget transfers, everyone who pays taxes in South Africa bears the burden. If it is funded by tolls, it is mostly residents of Gauteng who pay for it, either directly or via its impact on increases in the cost of living in the area. It might be seen as intuitively fairer to make the people in the region, who will benefit from the road, pay for that privilege. However, as with most tax policies, there is often a trade-off between the interests of fairness and those of efficiency.  In this case, what also needs to be taken into account is how expensive it is to collect these tolls.

Government can finance expenditures either through tax collections or debt issuance. In the 2010/11 financial year, SARS collected R 757.5 billion in taxation, and its operational expenditure amounted to R7.5 billion. This equates to a collections cost rate of approximately 1%. It is much more expensive to raise money through debt funding – for example, in January 2012, an issue of South African USD-denominated government debt was priced at an interest rate of 4.665%. 

In contrast, the tender to the firm which will collect the GFIP tolls for the first five years of operation is reported to be worth R6.2 billion – and this may not include all operating costs. Assuming that this cost is in addition to the R20 billion build cost for the GFIP, then the tolls collections cost rate works out to at least 31% (it is much higher if the build cost includes the operating costs). 

What does this imply for policy decision makers? If the GFIP had been paid for by the Treasury alone, then the cost of having SARS collect the money would be R200 million. If it had been funded by USD-denominated debt, it would have cost around R930 million. Regardless of the obvious efficiency savings from this approach, there are still good economic reasons to have Gauteng drivers and consumers carry a disproportionate share of the total project costs.  But not at any cost – and the range between the SARS collection rate and the debt financing cost should have been held out as an ideal benchmark.   Against this benchmark, the additional operational costs of R5-6 billion for the pleasure of using a highly sophisticated gantry system, seem extreme.  It would be interesting to compare the collection costs of the GFIP to those of other national and international infrastructure projects of this kind.

Much of the damage has clearly been done.   The roads are built, the gantries installed, and the Treasury has agreed to a partial bail-out of the project.  Can anything further be rescued from this mess?

What is not clear from the publicly available data is how many of these toll collection costs are already sunk, for example in building gantries, which cannot be recovered at this point. It seems likely however that some cost saving is still possible. Further investigation is needed before any additional funds are allocated towards the GFIP.  The gantries could remain in place as a reminder to all of the costs of getting the policy process so wrong.

Ideally, all these policy questions should have been decided in an open and transparent manner, well before the GFIP build program began, and before money had already been sunk into an inappropriate  and costly tolling system. Instead, the scope of policy action is limited by what has already been done, and by an absence of public data on what has actually been spent, and on what. An ideal means for conducting this analysis would have been a regulatory impact assessment.  For more on this, see the work done by DNA Economics for the Road Freight Association.