Performance and Expenditure Review Provincial Roads

Final Report February 2018

Research commissioned by the National Treasury

PER - Provincial Roads 2

Project Team:

Jonathan Carter Senior Economist with Cornerstone Economic Research [email protected]

Matthew Townshend Senior Economist with Cornerstone Economic Research [email protected]

Lungisani Ragwala Researcher with Cornerstone Economic Research [email protected]

Kobus Visser CEO at ERO-Engineers [email protected]

Roger Purchase CEO at tpa Consulting [email protected]

FEBRUARY 2018 PER – Provincial Roads i

Table of Contents LIST OF ABBREVIATIONS AND DEFINITIONS ...... 3 EXECUTIVE SUMMARY ...... 4 1. INTRODUCTION ...... 7 1.1 OVERVIEW OF THE PROVINCIAL ROADS NETWORK ...... 7 1.2 SNAPSHOT OF PROVINCIAL ROADS EXPENDITURE ...... 9 1.3 PURPOSE AND SCOPE ...... 10 2. INSTITUTIONAL ARRANGEMENTS IN THE PROVINCIAL ROADS SECTOR ...... 11 2.1 THE NATIONAL DEPARTMENT OF TRANSPORT ...... 11 2.1.1 Estimate national funding needs for provincial roads ...... 12 2.1.2 Technical norms and standards and indicator benchmarks ...... 12 2.1.3 Transferring officer of the Provincial Roads Maintenance Grant ...... 13 2.2 PROVINCIAL TRANSPORT DEPARTMENTS ...... 13 2.3 COMMITTEE OF LAND TRANSPORT OFFICIALS ...... 13 2.4 SANRAL ...... 13 2.5 CONSULTANTS AND CONTRACTORS ...... 14 3. PROVINCIAL ROADS FINANCING ...... 15 3.1 FUNDING SOURCES ...... 15 3.1.1 Funding of provincial roads in the intergovernmental system ...... 15 3.1.2 Discretionary funds ...... 15 3.1.3 The PRMG ...... 15 3.1.4 Budget ratios ...... 16 3.2 SUMMARY OF THE EXPENDITURE ANALYSIS ...... 16 3.3 THE FUNDING SHORTFALL IN THE SECTOR ...... 17 4. ROAD MODELS AND ANALYSIS ...... 19 4.1 BASELINE UNIT COSTS ...... 19 4.2 MAJOR COST DRIVERS ...... 21 4.3 ROAD NETWORK MODEL ...... 21 4.4 ROAD PROJECT MODEL ...... 24 4.5 FUNDING SCENARIOS ...... 27 5. PROVINCIAL ROAD PROJECT AND CONTRACT REVIEWS ...... 30 5.1 PER PROCESS AND FEEDBACK ...... 30 5.2 ROAD PROJECT PROCESS ...... 31 5.3 MONITORING ROAD CONDITION AND DATA COLLECTION ...... 32 5.4 PROJECT IDENTIFICATION AND NAMING ...... 32 5.5 APPOINTMENT OF CONSULTANTS ...... 33 5.6 TENDER RULES AND PROCESSES ...... 33 5.7 PROJECT DESIGN AND SPECIFICATIONS ...... 34 5.8 PROJECT WORK ...... 35 5.9 REFLECTING ON THE PROJECT PROCESS ...... 36 5.9.1 A mental map of the project process ...... 36 5.9.2 Skills and experience is crucial ...... 36 5.9.3 Activities that can be checked during oversight ...... 37 6. CONCLUSIONS ...... 38 6.1 THE ROAD NETWORK MODEL AND ROAD PROJECT MODEL ...... 38 6.2 ADDRESSING FUNDING SHORTFALLS IN THE ROADS SECTOR ...... 38 6.3 GAINING MORE FROM THE EXPENDITURE REPORTING SYSTEMS ...... 38 6.4 MINIMISING RISKS IN THE PROJECT CYCLE ...... 39 REFERENCES ...... 40

DRAFT OCTOBER 2017 PER – Provincial Roads ii

7. ANNEXURE 1 - WORKING PAPER 1: COST-EFFECTIVE PROVISION OF LOW-VOLUME ROADS IN SOUTH AFRICA 41 8. ANNEXURE 2 - WORKING PAPER 2: AN ECONOMICS-BASED ROAD CLASSIFICATION SYSTEM FOR SOUTH AFRICA ...... 52

DRAFT OCTOBER 2017 PER – Provincial Roads 3

List of abbreviations and definitions

BAS Basic Accounting System CIDB Construction Industry Development Board COLTO Committee of Land Transport Officials DoT Department of Transport IRM Infrastructure Reporting Model MinMEC Ministers and Members of Executive Councils PER Provincial Expenditure Review PES Provincial Equitable Share PMS Pavement Management Systems PRMG Provincial Roads Maintenance Grant RAMP Road Asset Management Plan RAMS Road Asset Management Systems RCB Roads Coordinating Body SANRAL South African National Roads Agency Limited SCOA Standard Chart of Accounts VCI Visual Condition Index

DRAFT FEBRUARY 2018 PER – Provincial Roads 4

Executive Summary

The South African road network consists of national, provincial and municipal roads. SANRAL, as an entity of the National Department of Transport (DoT), is the implementing agent of the national roads network and, along with the DoT, plays a key role in influencing policy and setting standards. Provincial roads are a provincial competence. Provincial road networks are maintained by the relevant provincial departments, although some roads have been transferred to SANRAL to be maintained by them. The provincial road network is a connector network, and plays a crucial role in supporting economic activity and providing access to social services within the provinces. The provincial road networks, as required by the Provincial Roads Maintenance Grant (PRMG), are assessed on a regular basis. The condition of provincial roads varies considerably, but the overall condition of networks is less than optimal. Paved road networks tend to be in a better state, with most roads in fair or good condition. The condition of the paved provincial road networks has, however, still experience deterioration. Most provincial roads are unpaved, and these are largely in poor or very poor condition. This presents a variety of challenges in an environment of constrained budgets. Firstly, provinces must decide which roads to prioritise, and, because they need to arrest further deterioration of the network and maintain high volume roads, roads in better condition often enjoy first claim on available resources. This best-first approach – which calls for roads in good condition to be maintained first, given that delays in required maintenance activities exponentially increases the total cost of works required – has meant that limited maintenance budgets are allocated repeatedly to the same set of high-volume roads, with low-volume routes often being neglected. Although neglecting the maintenance of some roads might be justifiable in a tight fiscal environment, these decisions must also take account of communities’ access needs and the need to promote of economic growth. Provinces fund their roads budgets from the provincial equitable share and the PRMG from the National DoT. Provinces spent R21.6 billion on their road network in 2015/16, of which R9.85 billion came from the PRMG. On aggregate, expenditure on roads takes up 4 per cent of total provincial expenditure. The PRMG supplements provincial budgets, and accounts for just over 20 per cent up to 80 per cent of provincial roads spending, depending on the province. The PRMG is allocated via the PRMG formula, which takes a variety of factors into account, such as the extent of the road network under a province’s management and the underlying condition of that network. When the grant was introduced, it was envisaged that the combination of factors in the formula would incentivise provinces to prioritise maintenance work, as this would ensure that roads budgets would purchase optimal value-for-money. Simplistically, this value for money was to be achieved by maintaining roads in good enough condition to prevent the need for more expensive works. Key to this is that grant could not be used for expenditure on new roads building or for upgrading of roads. One of the costing models developed through this PER will help plan PRMG allocations using network data from the provinces. Data drawn from the Basic Accounting System (BAS) presents expenditure in four key categories: repairs and maintenance, upgrade and additions, refurbishment and rehabilitation, and new infrastructure. By total expenditure, new infrastructure is the smallest, and its share has been declining. Between 2013/14 and 2015/16, on refurbishment and rehabilitation has been the largest expenditure item, followed closely by repairs and maintenance. Many roads projects involve a combination of these types of works, but an individual roads project can be classified against one category only. So even though provinces understand how works must be categorised, it not clear from the BAS data how accurate the expenditure totals per category are. There are variations in the ways provinces report expenditure against the segments in BAS, especially with regards to information recorded at different levels. Much richer expenditure

DRAFT FEBRUARY 2018 PER – Provincial Roads 5 information could be achieved if the National DoT, National Treasury and the provinces worked together to agree on uniform approaches to recording information in BAS, especially in the Asset, Project and Regional segments. Simple, easily implementable changes would yield rich expenditure data that could support valuable cross-province comparisons. Nevertheless, such changes to the Standard Chart of Accounts and BAS will not yield sufficiently disaggregated project data to support the unit cost analyses required to achieve the objectives of this PER. This was known before the project commenced. Therefore, much energy was invested in collecting project documents. Despite consultations and individual meetings with provinces only three provinces provided a full set of project documents. Provinces were generally reluctant to make project documentation available, which is worrying and was frustrating. Only two provinces could provide electronic copies of project documents during the time period, which is of concern and, inadvertently, an important finding of the study. The following mental map identifies the core elements and events in the project process, shaded in black. The grey cells identify resources, events or processes that can address key risks at each stage in the project process. It should be read clockwise from the red arrow.

Key sources of risks identified during the review of the project documents are:  Projects are incorrectly specified resulting in project designs that are not fit for purpose creating inefficiencies.  Selected bids lack balance and as they are not correctly evaluated create substantial financial risks which the provinces are contractually locked into.  Tenders are awarded to companies that lack adequate experience and skills because evaluation processes that lack sufficient rigour. These risks can largely be mitigated if the workforce in the department is skilled, experienced and conduct their work with a high degree of professionalism. More needs to be done by provinces to build the capacity of their roads departments. Importantly they should be offering more bursaries to engineering students so that a sufficiently large pipeline of experienced engineers becomes available. In the short term greater use of uniform tender documents and standardised project specifications will help to reduce errors.

DRAFT FEBRUARY 2018 PER – Provincial Roads 6

Two costing models have been developed: a network model and a project cost model. The network model can be used by role players and individual provinces to estimate the province-specific and total maintenance backlogs in the sector. The network model provides key network characteristics, including the disaggregation of road classes within provinces and the underlying road conditions. This data is used to estimate the maintenance backlog in each province and for the entire sector. The available data suggests that the current maintenance backlog for all provincial paved roads and gravel roads is R50.7 billion and R65.5 billion, respectively. The roads project model is intended to be used as frameworks to help officials ensure the correct specification for each project, and to set a par value against which cost estimates can be tested. It has the potential create a knowledge base that will help provinces address the above-mentioned risks in the project process by  estimating the effect that specific variables have on the cost of road projects;  interrogate the project specifications and costs generated by consultants;  analyse available road surface options and select the most cost-effective alternative; and  record road project information and thereby develop their own functional cost databases that informs the above three listed processes.

DRAFT FEBRUARY 2018 PER – Provincial Roads 7

1. Introduction

Provincial roads are a vital cog in the South African economy, and key to ensuring access to social services. They are generally a high-use infrastructure, the condition of which has a significant impact on users’ experience and their potential contribution to economic activity. The need to maintain roads does not wait for provincial road management systems to be in place or for budget processes to run their course, with the costs of maintenance rising exponentially with the deterioration of their condition. Furthermore, what happens in one section of the network can dramatically affect other parts. If a stretch of road goes into disrepair, fewer people use that section, which has an impact on alternate routes and feeder roads. At any one time, there are many sections of road in a network that require work, but resource constraints mean that only a small proportion can be selected for work to be done. This document is a report on a performance expenditure review (PER) on provincial roads. The key output of this PER is a costing model that enables users to develop cost estimates of a variety of kinds of roads projects in a variety of different conditions. The costing model can be applied and adapted to all conditions that exist in South Africa if the relevant data is provided and input. The model can generate estimates for all conditions, though some judgement will be required if it is to be used optimally. Users can utilise the model to obtain alternate costings of projects by tweaking variables, and therefore have a powerful tool to quickly compare different approaches to the same sections of roads. This also supports institutional knowledge accumulation and learning as officials can quickly compare the impact of these differences on project costs. The project originally aimed to collect project documents from provinces and to use these as a source of data for the costing model. Given the objectives of the PER, it was expected that provinces would be willing to share project documents with the project team. Unfortunately, only a few provinces were fully cooperative, making an effort to provide the project team with project documents. A notable finding of the project is the apparent inability of most provinces to provide electronic copies of project documents due to a lack of a digital project repository. There is no sensible reason why this is the case. In the absence of historic project costs, the model is underpinned by technical information provided by engineers, and sourced from a database of a range of roads projects implemented across South Africa, which is, in any event, probably a more accurate source of unit costs than the project documents would have been. The absence of project data for most provinces means it is not possible to assess the efficiency and cost-effectiveness of provinces’ road maintenance programmes. The model is discussed in detail in Section 4. 1.1 Overview of the provincial roads network South Africa’s road network, including unproclaimed roads, is approximately 750 000km. This is the tenth longest road network in the world. Responsibility for it is split between the national, provincial, and local government road authorities. The provincial road network is approximately 275 000km, about 45 per cent of the total proclaimed road network. Travel on South Africa’s roads paved runs to about 32 billion vehicle-kms per year. This would include most travel on the 47 000km of paved provincial roads. More than 80 per cent of the total provincial road network, however, consists of low-volume gravel roads. These gravel roads tend to be in rural and peri-urban regions, and provide isolated communities with access to public services, economic centres and other key facilities. Provinces are responsible for the development and management of all provincial roads, unless a road has been explicitly transferred to another agency. Figure 1 illustrates the nine provincial road networks, disaggregated by paved and gravel roads. While the extent of the paved road networks is similar across provinces, there are significant differences in the length of gravel roads. Importantly, network length is not necessarily a function of economic activity. Provinces with relatively small economies (Eastern Cape, Free State and Northern Cape) manage some of the

DRAFT FEBRUARY 2018 PER – Provincial Roads 8 largest road networks as a result of their geographic area and population density, and as the importance of farming to the provincial economy. Figure 1 Provincial road networks, 2017

Source: Provincial Road Asset Management Plans Provincial road authorities have allowed the condition of these important assets to decline. Figure 2 and Figure 3 detail the standardised Visual Condition Index (VCI) of the paved and unpaved provincial networks. Roughly two-thirds of gravel roads under provincial management are in poor or very poor condition (Figure 3), and, while relatively fewer paved roads are in poor or very poor condition (Figure 2), the paved network is still below acceptable operational standards.1 This outcome is attributed to a range of factors, including limited human resources to manage the extensive road networks and constrained financial resources, especially in provinces with smaller economies. Considering the importance of the provincial road networks and the tight resource environment, improvements should be sought in the regulatory environment, provincial operations and sector oversight to realise improved value-for-money in the sector.

1 No more than 10 per cent of roads can be in poor or very poor condition (reference to come).

DRAFT FEBRUARY 2018 PER – Provincial Roads 9

Figure 2 Condition of paved provincial roads

Source: Provincial Road Asset Management Plans Figure 3 Condition of unpaved provincial roads

Source: Provincial Road Asset Management Plans 1.2 Snapshot of provincial roads expenditure Table 1 provides a snapshot of provincial roads expenditure between 2012/13 and 2015/16.

DRAFT FEBRUARY 2018 PER – Provincial Roads 10

Table 1 Snapshot of provincial roads expenditure

Roads/Total PRMG/Roads Provincial Roads Annual Average Expenditure R 000s Expenditure Change 2013/14 - PRMG 2015/16 Expenditure 2015/16 2015/16 Average 2013/14 to 2015/16 2013/14 2015/16 Eastern Cape 2,001,953 5% 1,289,151 3% 70% 62% Free State 1,518,314 5% 1,142,796 5% 79% 75% Gauteng 2,177,112 7% 455,821 2% 39% 21% KwaZulu Natal 6,977,023 5% 1,779,255 7% 25% 22% Limpopo 1,642,199 2% 994,762 2% 49% 61% Mpumalanga 2,314,671 5% 1,719,418 6% 73% 74% North West 1,335,530 -1% 788,060 10% 47% 62% Northern Cape 1,066,517 8% 822,430 3% 89% 74% Western Cape 2,607,862 16% 858,962 5% 26% 33% Total/Average 21,641,181 6% 9,850,655 4% 47% 45% In 2015/16, province’s spent R21.6 billion on roads. On aggregate, this was a 6 per cent average annual increase from 2013/14, when provinces spent a total of R18.4 billion. A key source of funding for provincial roads is the Provincial Roads Maintenance Grant (PRMG), which was R9.6 billion in 2015/16, up from R8.7 billion in 2013/14. Depending on the province, the PRMG accounts for between 21 and 89 per cent of expenditure. It contributed to 47 per cent of total provincial roads expenditure in 2013/14, and this value decreased to 45 per cent in 2015/16. On aggregate, provinces spent 4 per cent of their total provincial budgets on roads, though this ranged from 2 to 10 per cent. 1.3 Purpose and scope This PER aims to: (a) quantify and analyse expenditure by provincial governments on the construction and maintenance of provincial roads between 2013/14 and 2015/16; (b) assess value-for-money by identifying relevant metrics for comparative expenditure across provinces based on project/road specifications; (c) determine key drivers of expenditure and expenditure growth on provincial road construction and maintenance; (d) make recommendations for achieving greater value-for-money. The scope of the project is exclusively limited to expenditure on provincial roads. This document is the report on this PER. Outputs that have been delivered through this project include:  a log frame in Microsoft Excel;  a Microsoft PowerPoint presentation with an institutional analysis and a flow of funds analysis;  a Microsoft Word document containing the expenditure review; and  a costing model developed in Microsoft Excel. This contains both a project and a network model. As far as is necessary, elements of the above outputs have been replicated in this report.

DRAFT FEBRUARY 2018 PER – Provincial Roads 11

2. Institutional Arrangements in the Provincial Roads Sector

The institutional arrangements in the provincial roads sector, and the activities and performance of the relevant stakeholders, frame this PER. , There are six key stakeholder groups that directly affect expenditure outcomes within the sector (Figure 4): national DoT; provincial road authorities; the South African National Roads Agency (SANRAL); contractors and consultants; the Roads Coordinating Body (RCB); and industry bodies such as the Construction Industry Development Board (CIDB) and the Committee of Land Transport Officials (COLTO). Prior to analysing historical provincial road expenditure, we will present the findings of this review exercise, surveys conducted with staff within the provincial road departments, and the experience of three senior pavement engineers to explore the roles and responsibilities of the stakeholder groups. This section briefly explores weaknesses in the activities of other stakeholders that potentially have a material effect on expenditure outcomes in the sector. This approach helps determine whether structural limitations to value-for-money exist, such as policy gaps or missing specifications. The discussion allows for constructive and targeted engagement with the challenges in the sector. Figure 4 Stakeholders in the provincial roads sector

2.1 The National Department of Transport The national DoT plays a largely facilitative and regulatory role in the provincial roads sector. It is responsible for transferring funds to provinces, monitoring and evaluation, general financial and non-financial oversight, and developing the policies, legislative framework, strategies, plans, regulations, and guidelines to assist the provincial roads authorities in carrying out their duties effectively. Provincial road sector policies Whilst policy direction in the sector has generally been clear,2 there are some specific areas in which provincial authorities would benefit from greater clarity. The first of these, which is topical amongst several organisations including the World Bank and the International Road Federation, is the need for a cost-effective surfacing policy for the approximately 160 000km of low-volume gravel roads under provincial control. Many provinces lack the necessary resources to maintain their entire low-volume road network, but such a policy would ensure that provinces are able to use their limited

2 The main transport policies are detailed in the 1996 White Paper on National Transport Policy and the Moving South Africa document.

DRAFT FEBRUARY 2018 PER – Provincial Roads 12 resources efficiently increase value-for-money. Other areas where greater policy clarity is desirable include:  A consistent methodology for prioritising road expenditure. As determined through a survey conducted with key provincial road officials, there is no consistent policy to guide how provinces set their maintenance schedules. This leads to inconsistent outcomes, increases the risk that limited road resources are not utilised efficiently.  A consistent methodology to estimate the economic significance of roads within the provincial networks. The survey indicated that provinces use varying sources of information to determine the economic contribution and importance of roads. None of these are sufficiently robust to ensure that the network is able to deliver on its social and economic purposes. The current set of analytical tools are biased to engineering indicators, are subjective, consider a limited number of economic variables, or are useful only at a project but not a network level. Limited resources mean that prioritisation of road expenditure is happening within all the provinces, but there is currently no methodology to guide how these maintenance schedules are set, leading to significant differences in approach. The prioritisation practices range from detailed assessments of traffic volumes and vehicle types to a crude combination of each road’s classification and VCI score. The likely outcome is that the available road budgets, especially in the provinces with less developed Road Asset Management Systems (RAMS), are not realising value-for-money. Moreover, provinces lack a system for considering how roads promote access to economic and social amenities, factors that play no role in the RAMSs. These basic access features include connecting isolated communities to public services such as education and healthcare. Despite the key socio-economic function of such roads, their maintenance is often overlooked given their low traffic volumes. Some provinces have piloted assigning scores to roads based on the number and type of facilities to which it is connected, but these pilots have not been rigorous or integrated with the existing RAMS data. The maintenance of these vital roads cannot be left to discretion or unreliable scoring processes, and therefore a decision should be reached on a uniform methodology to incorporate the basic access functions of each road into the RAMS. As part of this expenditure review, the following working papers have been included in the Annexures to guide these important policy discussions if/when they are initiated:  A cost-effective surfacing policy for low-volume roads in South Africa (Annexure 1)  Economic classification of the South Africa road network (Annexure 2)  Identification and optimisation of the Basic Access Road Network in South Africa (Annexure 3)

2.1.1 Estimate national funding needs for provincial roads The national DoT is responsible for estimating the national funding needs for provincial roads; bidding for funds for the PRMG on behalf of provinces; and negotiating with National Treasury on the conditions and structure of the PRMG through grant frameworks in the Division of Revenue Bill. These estimates should be based on the extent of the provincial roads maintenance backlog, which should be calculated by the provinces. These values, together with available budget and practical considerations about competing government priorities, inform the PRMG allocation criteria.

2.1.2 Technical norms and standards and indicator benchmarks The national DoT develops the technical standards and benchmarks with which provinces should comply, such as collecting and collating RAMS data and reporting processes. The RAMS data is in line with good engineering practices, but lacks key economic indicators – such as the value added

DRAFT FEBRUARY 2018 PER – Provincial Roads 13 by each road, the type of sectors a road supports, and whether a road is connected to basic service facilities such as schools or hospitals. From an efficiency perspective, it is imperative that the expenditure and performance reporting standards are applicable to all provinces, and that provinces adopt and abide by them. During the review process, there were instances of selective application of standards by provinces, such as reporting standards, and loose compliance with indicator benchmarks. One challenges is that it is not clear how compliant provinces might be rewarded or how non-compliance should be addressed. This is because provincial roads are listed in Schedule 5A of the Constitution as an exclusive provincial legislative competence. Therefore, the national DoT cannot prescribe minimum norms and standards to provinces.

2.1.3 Transferring officer of the Provincial Roads Maintenance Grant The national DoT is the transferring national officer of the PRMG, and is therefore responsible for ensuring provinces comply with the conditions of the grant and for reporting to National Treasury on the financial and non-financial performance of the grant. Importantly, the national DoT has initiated moves to alter the conditions of the PRMG to allow a portion of the grant to be allocated to upgrading gravel roads. If the objectives and affordability of this decision can be clearly determined, this is something this report endorses as it finds that all gravel roads worth maintaining should be sealed at as fast a rate as possible within budget limits. 2.2 Provincial transport departments Some provincial roads departments are a part of a department that includes Public Works. Limpopo’s provincial network is managed by a combination of the roads department and the Limpopo Roads Agency. As custodians of their road networks, provinces are responsible for: setting policy priorities, allocating budgets across these priorities to achieve maximum value-for-money, managing the implementation of projects themselves or outsourcing this to engineering companies, and collecting and maintaining data on their road networks for planning and reporting purposes. 2.3 Committee of Land Transport Officials COLTO, who is the de facto custodians of the knowledge base for the road construction industry, provides technical support to the RCB. Amongst other outputs, this support consists of various technical guidelines, recommendations and specification reports that inform how road works should be conducted. COLTO consists of technical sub-committees that make necessary recommendations to the MinMEC. The documents produced by COLTO generally take two forms, Technical Manuals for Highways or Technical Recommendations for Highways. As core technical references, these documents should be more strictly adhered to by provincial road authorities. This outcome has tried to be enforced by making compliance with the specific standards and methods part of the PRMG conditions, but there are issues with minoring compliance by provinces and holding them to account. Better enforcement would help ensure that a consistent standard is realised for road works, project reporting, project specifications (consultants will be guided in the development of project documents) and tenders (contractors will have set documents to refer to). Where deviations from these documents occur, provinces should provide sufficient justifications. 2.4 SANRAL SANRAL is a key role player with respect to South African road networks. It is an agency of the national DoT, and acts as a consultant for provinces, manages sections of provincial road networks under contract from provinces, sits on the RCB and COLTO, and reviews policy documents from the national DoT and provides advice and recommendations on these.

DRAFT FEBRUARY 2018 PER – Provincial Roads 14

SANRAL currently manages several provincial roads, mostly in the Eastern Cape, Free State, and Gauteng. Roads are generally transferred to SANRAL when a province does not have the capacity to provide the intensive maintenance activities required by the high traffic volumes on these roads. As funding doesn’t automatically follow the transferred assets, it is not in the provinces’ financial interests to maintain control of these high-volume or distressed sections of road. While SANRAL must maintain the roads using existing budget envelopes in the first year of ownership, additional funding can be applied for or generated via tolls in subsequent years. This arrangement may create perverse incentives, whereby provinces allow the condition of certain roads to decline knowing that SANRAL – who is sometimes accused of mission creep - is motivated to absorb them within their network for financial and prestige purposes. 2.5 Consultants and contractors Provinces use the services of consultants and contractors for a range of purposes. Consultants generally develop the project documents that are put out as part of the tender advertisements, which then guide the work of contractors. Contactors conduct the road works, under the supervision of a province-employed engineer, and also manage and update many of the RAMS. Contractors perform two roles with regard to roads projects. Firstly, projects are outsourced to engineering firms. When this happens, the engineering company designs the project, manages the procurement process and oversees the contractor. This is described as an “outsourced” project. The engineering firm takes responsibility for ensuring the project is correctly implemented. Secondly, the provincial department can design the project, manage the procurement and take responsibility for project management. This is described as a “contracted” project.

DRAFT FEBRUARY 2018 PER – Provincial Roads 15

3. Provincial Roads Financing

3.1 Funding sources Province roads authorities receive a portion of the provincial equitable share through the provincial budget process, an amount that is supplemented with the PRMG as well as a small amount of own revenues. At present, the PRMG can only be allocated to the maintenance and repair of roads (not new construction or upgrading). This limitation does not apply to other funds.

3.1.1 Funding of provincial roads in the intergovernmental system The national DoT has a responsibility to bid for funds for the PRMG on behalf of provinces during the budget process, and efforts in this regard are focused on the value of allocations to the PRMG. The national DoT could make arguments for increasing allocations to the provincial equitable share so provinces have more funding for roads. Purely from a roads perspective, this would be ill-advised as there is no direct way for national government to ringfence provincial equitable share allocations for roads. Because provincial roads are not a concurrent (Schedule 4) function, the national DoT is not even entitled to prescribe the minimum standards of road condition that all provincial roads should be maintained at. Instead, provincial roads are a Schedule 5A function, which means that it is an exclusive provincial legislative competence. While the national DoT may set policy, identify strategic goals for the sector and issue guidelines that contain norms and standards, the DoT cannot enforce these. The national DoT, however, can and does make compliance with and/or use of specific industry-related standards a condition for the disbursement of the PRMG. However, because the PRMG is a supplementary grant, its conditions cannot be applied to spending funded through the equitable share, which is mixed with the PRMG. It is crucial that the funding for provincial roads rewards those provinces that keep their networks in good condition by preventing sections of their networks from degrading into states that require disproportionately large spending to achieve acceptable conditions. If funding is allocated on the basis of the need for expenditure according to actual condition of the network, then provinces have an implicit incentive to allow roads to degrade to poorer condition, as this would result in larger allocation of funds. That roads are funded through a combination of own (discretionary) and supplementary grant funding gives provinces the flexibility to balance different types of roads expenditure so as to maintain their networks in the best possible condition given constrained resources, while also allowing the national DoT to influence key strategic decisions.

3.1.2 Discretionary funds Discretionary funds consist of funding from the provincial equitable share and provinces’ own revenues, including car licence fees. The formula that determines provincial equitable shares does not include an explicit variable for the length, condition or economic importance of provincial roads, though economic activity does affect provinces’ shares.

3.1.3 The PRMG The PRMG is a Schedule 4, or supplementary, conditional grant. This means provinces should use it to supplement existing allocations to roads by the province. Although there is no required amount provinces should fund from their own sources, it would be reasonably expected that provinces would use the supplementary grant for a maximum of around 40 per cent of total expenditure. In fact, the PRMG accounts for between 21 per cent in Gauteng and 89 per cent in Northern Cape. The allocation of the grant is conditional on the provinces completing and tabling a Road Asset Management Plan (RAMP), collecting road condition data, and the submission of monthly reports on the Infrastructure Reporting Model. It is also dependent on the submission of project plans.

DRAFT FEBRUARY 2018 PER – Provincial Roads 16

The PRMG was created in 2011, when the roads portion of the Infrastructure Grant to Provinces was unbundled from the health and education infrastructure funding. The emphasis of the grant has been on routine maintenance and maintenance and repair work. The intention behind this is that grant funds be used as cost-effectively as possible, broadly measured in terms of length of the road network maintained in fair to good condition. This focus should ensure asset preservation of the provincial road network. Provinces have not been permitted to use the grant for building new roads or upgrading existing roads. The construction of new roads creates an additional maintenance burden, which is something the provinces cannot handle financially, given the existing maintenance backlogs in provincial roads.

3.1.4 Budget ratios Table 2 uses the provincial road network lengths to convert the total road expenditure per province to a per km value. Because each road does not require the same amount of maintenance and not every road receives maintenance action every year, these ratios do not reflect how much provinces actually spend on each km of road and care should be taken before making comparisons. The figures do, however, give an indication of each department’s financial capacity relative to their managed networks. The data suggests Limpopo, North West, Eastern Cape, Northern Cape, and the Free State are severely under resourced – lacking even the R100 000 per annum estimated as necessary for the routine maintenance of 1 km of road. Table 2 Total road expenditure as a per km ratio

2013/14 2014/15 2015/6 2016/17 Gauteng R 331 837 R 415 550 R 431 590 R 407 996 KwaZulu Natal R 183 588 R 209 040 R 213 805 R 214 988 Western Cape R 117 244 R 123 255 R 153 295 R 183 094 Mpumalanga R 147 867 R 171 051 R 167 068 R 173 619 Limpopo R 83 892 R 74 645 R 87 184 North West R 75 761 R 62 594 R 67 502 R 73 914 Eastern Cape R 43 239 R 39 407 R 48 533 R 50 718 Northern Cape R 32 412 R 29 794 R 38 484 R 40 676 Free State R 31 605 R 34 389 R 33 355 R 36 311 3.2 Summary of the expenditure analysis In line with the PER methodology, expenditure data from the National Treasury dataset of provincial budgets for 2016/17 and 2017/18 were used to source data on roads expenditure between 2013/14 and 2016/17. The full expenditure analysis is included as a separate attachment to this report. Data from the Basic Accounting System (BAS) was used to analyse expenditure from 2013/14 to 2015/16. Expenditure on roads as a proportion on total provincial spending ranged from 2.3 per cent in Gauteng to 9.3 per cent in North West in 2015/16. On aggregate, expenditure on roads accounted for 4.4 per cent of total provincial expenditure in 2015/16. Annual average growth on roads expenditure between 2013/14 and 2016/17 was lower than growth on compensation of employees in all provinces except the Western Cape, and lower than growth in total provincial expenditure in all provinces except the Eastern Cape, Northern Cape and Western Cape. This suggests that expenditure on compensation of employees is squeezing the roads budgets in eight of the provinces. Furthermore, other expenditures were prioritised above roads expenditure in six of the provinces. The reliance on the PRMG as a source of funding varies substantially across provinces. There appear to be three groups, with the following proportions of the roads budget funded by the PRMG:  The big three: KwaZulu-Natal, Gauteng and Western Cape, between 26 and 30 per cent.  Far North: North West and Limpopo, between 55 and 56 per cent.

DRAFT FEBRUARY 2018 PER – Provincial Roads 17

 Other Rural: Eastern Cape, Mpumalanga, Free State and Northern Cape, between 68 and 80 per cent. BAS enables analysis of the following types of road expenditure:  Maintenance and repair  New Infrastructure  Rehabilitation and Refurbishment  Upgrades and Additions Provinces are unable to follow these expenditure categories absolutely. Provinces have been given guidance on the different terms used so there is no confusion as to how works should be classified. However, if a project includes components that fit in each type, the total project cost can only be recorded against one project type. The analysis of expenditure according to these categories must be viewed with this in mind. The analysis of expenditure by the above types of expenditure showed:  Expenditure on New Infrastructure makes up a small and falling share of spending.  On aggregate, Rehabilitation and Refurbishment accounted for most spending, followed by Maintenance and Repair and then Upgrades and Additions.  The detailed analysis of the segments shows that there is some variation in how provinces use and record information in BAS, limiting the value of cross-province analysis of these.  All the provinces implement most of their projects as outsourced projects (where an engineering company project-manages the contractors) except for KwaZulu-Natal, which runs all routine maintenance operations internally with the assistance from Area Consultants while outsourcing rehabilitation, refurbishment, and new infrastructure. Eastern Cape and Northern Cape project-manage a small proportion of their projects. Key points emerging from the analysis in the context of this PER are:  Provincial roads expenditure competes with education, health and social development for provincial equitable share allocations. Expenditure on roads is small in comparison to the expenditure on the other services.  There is wide variation in the relative proportion the PRMG makes up of total roads expenditure across provinces. It is notable that the PRMG makes up the smallest proportion of expenditure in the three provinces with the biggest economies. This is possibly a reflection of how the sizes of economies affect the political economies in the provinces and how they prioritise resources.  The existing reporting requirements in SCOA do not create expenditure data in BAS that can enable the detailed unit cost analysis of provincial roads expenditure that was sought by this PER. It is unlikely this will ever be the case, given the complexity of roads projects and the in-built limitations of how expenditure is reported in SCOA.  Despite the above, it is recommended that the National DoT, working with National Treasury, provide guidance to the provinces on reporting the SCOA segments, as there is potential to improve the usefulness of expenditure data in BAS. One of the primary goals should be to build a knowledge base which is accessible to all partners. 3.3 The funding shortfall in the sector Based on 2013 VCI data, SANRAL (Kannemeyer, 2014) estimated that R128 billion was required to strengthen paved provincial roads that were in poor to very poor condition. SANRAL also estimated that a further R19 billion was required to regravel all unpaved provincial roads that were in poor to very poor condition. Since 2013, the overall condition of the provincial paved road

DRAFT FEBRUARY 2018 PER – Provincial Roads 18 networks has improved, and some roads have been transferred from provinces. We therefore estimate that the total backlog for this portion of roads has fallen to around R50.7 billion. To estimate road maintenance backlogs, we applied the requisite remedial action to roads in very poor (major rehabilitation), poor (major rehabilitation), fair (minor rehabilitation), good (periodic maintenance), or very good condition (routine maintenance). An average cost for these activities based on the surface type was used. The condition of gravel road networks has deteriorated further as limited resources have meant that the bulk of this network has continued to be neglected. Moreover, it was found that the average per km regravelling cost captured by provinces in their RAMS (sometimes as low as R500 000) was a conservative estimate when compared with the actual cost of these projects reported in the project documents and by contractors (often around R800 000). We therefore estimate that the overall maintenance backlog for the provincial gravel road networks could be as high as R65.5 billion. Figure 5 maps the 7 781 provincial road projects that were loaded on the Infrastructure Reporting Model (IRM) for the period 2015/16–2017/18. The low number of dots is an artefact of the scale used and of instances of overlapping projects. In addition, a single point rather than the entire length of the project was used to log the projects. Projects prior to this period were excluded from this exercise, as no GPS coordinates were reported. The findings suggest that provinces generally conduct regular maintenance on the same sets of roads. While this is sensible in the context of budget constraints, with the high-volume mobility roads being given maintenance priority, one effect is the emergence of areas, most notably rural regions, in which roads are consistently overlooked for maintenance. If funding in the sector were to become more constrained, the concentration of projects along major nodes may well be exacerbated. These 7 781 projects, which were all the projects recorded on the IRM for that period, are possibly not the full set of provincial road projects as there are instances of provinces packaging maintenance projects for procurement reasons and bundling them as a single IRM entry. As a result, provinces do not capture each individual maintenance project or works, with each projects GPS data, in the IRM. Provinces followed the incorrect process in these instances, something that should be addressed by the national DoT. Figure 5 Road maintenance shadows

Source: Infrastructure Reporting Model

DRAFT FEBRUARY 2018 PER – Provincial Roads 19

4. Road Models and Analysis

Two costing models have been developed: a Network Cost Model and a Project Cost Model. These models are intended to be user-friendly tools that enable stakeholders in the provincial roads sector to:  estimate the provincial and total maintenance backlogs in the sector, disaggregated by surfaced, gravel, and strategic road networks. This will allow National Treasury and the national DoT to validate the provincial RAMS data and any external estimates of road maintenance backlogs. The Network Cost Model provides useful information to help plan PRMG allocations over the MTEF.  estimate the effect that specific variables have on the cost of road projects.  interrogate more effectively project specifications and costs generated by consultants. Officials can use this tool to cross-check the inputs provided by consultants. Where a deviation between the expected and quoted costs is present, the tool can identify issues as a basis for targeted questions to understand the underlying reasons for this. This will also help provinces limit instances of under- and over-specification of road projects.  analyse available road surface options and select the most cost-effective alternative.  record road project information and thereby develop their own functional cost databases.  Help project managers report on the planned and actual expenditure on specified types of road work on all projects. 4.1 Baseline unit costs Unit cost data informs both the Network and Project Cost Models. The model takes a bottoms-up approach to building these estimates because of the range of variable in play, including: price volatility and the multiple road surfaces,3 road work activities,4 wide differences in initial road condition, traffic volumes,5 climatic conditions,6 contractor types, resource availability (i.e. the proximity of suitable building materials), community agreements, and labour-intensity of works that were modelled. To construct the database entirely from project data would have been problematic for three main reasons: it was not possible to acquire the volume of project data necessary to generate a robust sample across all the above project scenarios; many of the collected project documents captured a range of activities that could not be disaggregated retrospectively; and much of the actual project data already captured some level of pricing distortions. The following 10-step methodology was therefore applied to generate the baseline unit cost data: Step 1: Estimate an efficient cost for each activity using first principles. These are undistorted input prices, efficient but sufficient quantities of materials, and reasonable timeframes and workload distributions. Step 2: Cross-check the baseline cost data with actual data from relevant recent projects and tender proposals that were known to be competitive. These reference costs were based

3 Gravel; Slurry seal; Sand seal; 14 mm Cape Seal with 1 slurry; 14 + 7 mm Double Seal; Geotextile seal; Split seal; Choked seal; Inverted seal; Segmented block paving; 30 mm Hot Mix Asphalt (HMA); 40 mm HMA; 50mm HMA; and ultra-thin reinforced concrete pavement. 4 Construction; Routine maintenance; Periodic Maintenance; Minor rehabilitation; Major rehabilitation. 5 Low volume roads (Annual Average Daily Traffic (AADT) < 219); Low-medium volume roads (220 < AADT > 499); Medium volume roads (500 < AADT > 999); Medium-high volume roads (1000 < AADT > 2999); and high volume roads (AADT > 3000). 6 Standard profile; High moisture content (>20 Thorthwaite); High precipitation volumes; Steep gradients (>8%); Rolling hills; and any combination of these.

DRAFT FEBRUARY 2018 PER – Provincial Roads 20

on winning tenders and projects that matched the set characteristics of the specific project types (surface type, climatic conditions, traffic volumes, activity). Step 3: Cross-check the baseline cost data with reliable supplier databases. This was done using market data and historical price datasheets developed by members of the project team. Step 4: Have the baseline cost data cross-checked by three experienced pavement engineers. Step 5: Have the baseline cost data cross-checked by key industry bodies, such as the Southern African Bitumen Association and SANRAL officials. Step 6: Compare the baseline cost data with the cost estimates presented in the RAMS. Step 7: Cross-check the findings generated using the baseline cost data with findings from other countries (Namibia, India and Australia). Step 8: Submit the baseline cost data and methodology to the International Road Federation for peer review. Step 9: Present the baseline cost data and methodology at a high-level conference – The 2017 World Road Meeting – for further peer review and scrutiny. Step 10: Create the necessary functionality within the models so that provinces can adapt the cost data if necessary and/or begin to generate their own databases.

DRAFT FEBRUARY 2018 PER – Provincial Roads 21

4.2 Major cost drivers

The major drivers that affect the cost of road works are discussed in Table 3. Table 3 Cost drivers in road construction

Cost Driver How this affects costs Surface type Each surface type is comprised of different quantities, ratios, and types of inputs. Road condition and the type of distresses affect the type and magnitude of required remedial Road condition works. It can cost as much as seven-times the routine maintenance cost to repair a road in poor condition. Traffic volumes determine the required pavement structure, type of road surface, frequency of Traffic volume maintenance actions and design life of the road. Heavier vehicles impose higher stresses on roads, thereby requiring stronger pavement structures and road surfaces, and also relatively more frequent maintenance activities. Special Traffic type dispensation has been made within the PRMG for the Mpumalanga coal routes for this exact reason. Gradients steeper than 8:100 require added paved side drains and segmented block paving or concrete to cope with the steep slope. This generally increases the cost of a surfaced road Gradient by approximately 20%. Extra passing lanes are also sometimes required. Gravel roads should also receive an additional two blading/grading events per annum. High moisture content requires a layer of rock fill to stabilise the road foundation and subsoil Moisture content drains. This increases the cost of a surfaced road by approximately 5-7%. High rainfall volumes require additional pavement strengthening and drainage structures. This Rainfall volume increases the cost of a road project by approximately 7%. Gravel roads should also receive an additional two blading/grading events per annum. There are various sizes of contractors, ranging from small- to large-scale. The current tendering system includes a 20% BBBEE contribution to the evaluation of tenders. This allows Contractor type small-scale contractors a marginal price advantage. Supervision costs are higher when small- scale contractors are used. Resource availability, especially local materials, differs depending on regions. For instance, Availability of the distance to borrow pits can vary widely between projects, with the average distance resources tending to have increased since revised regulations on opening new borrow pits was passed by the Department of Environmental Affairs in 2002. Projects are done using some combination of traditional and labour-intensive construction Labour-intensity of methods. Labour-intensive construction tends to extend the duration of projects as productivity works rates are relatively lower, but can be a cheaper option depending on the EPWP/low-skilled wage rate. Community Community agreements add a surcharge to road projects. The amount differs based on the agreements community in question and the type of agreement reached. The price of many key inputs is volatile, which has a distortionary effect on the cost of road works. For instance, the price of bitumen spiked in the mid-2000s, and again in 2013, because Input inflation of supply shortages. The price of fuel, which like bitumen also follows movements in the international oil price, is another unpredictable input.

4.3 Road Network Model As part of their RAMS submissions to the national DoT, each province applies a “standard” costs to different road maintenance works based on the surface type and VCI of each road. There is no system to determine these standard values, and as such there are variations in the accuracy of the unit costs and the frequency at which they are updated. These standard costs are applied to the corresponding sections of the road network (matched according to surface type and conditions) and aggregated to estimate a total road maintenance backlog. While these estimates may on

DRAFT FEBRUARY 2018 PER – Provincial Roads 22 aggregate approximate the actual road maintenance backlog, guidelines are required to monitor or regulate the standard costs that provinces apply to groups of road works and surface types. The costing models presented below were developed to assist with this task. The Road Network Model, shown in Figure 6, estimates the funding required to maintain the provincial road networks. The analysis is disaggregated by the paved, unpaved, and strategic road networks in each province. The model creates graphs of the selected provincial network to support easy analysis. Examples of graphs created of the Western Cape road network are shown in Figure 7 and Figure 8. Users have the option to specify their network profiles and input what percentage of their network is standard, steep, rolling, or in high-moisture areas (see the blue cells in Figure 6). The functional road classifications and VCI are also captured and factored into the estimated maintenance costs. Roads in poor and very poor condition are set to receive major rehabilitation, roads in fair condition to receive minor rehabilitation, and roads in good condition to receive periodic and routine maintenance. This functionality adds the necessary detail to the estimated maintenance costs.

Figure 6 Snapshot of the Road Network Model

DRAFT FEBRUARY 2018 PER – Provincial Roads 23

Figure 7 Graphs of the road network created by the road network model, Example 1

Figure 8 Graphs of the road network created by the road network model, Example 2

The model helps to standardise the approach to estimating maintenance backlog costs, and where there are deviations between this and other methods, it enables the user to test for the source of the difference. The estimated maintenance backlogs were compared with the reported maintenance backlogs in North West. This was the only province that submitted RAMS data in a format that could be used. The estimates compared well for the paved road network, with the model indicating an approximate maintenance backlog of R5.6 billion compared with the province’s estimate of R6.4 billion. The average cost of R500 000 per km used by the North West in their RAMS for rehabilitation of gravel roads in poor and very poor condition appears a conservative estimate, with such projects often currently costing around R800 000. The result is that the maintenance backlog for gravel roads in the province could be as high as R11.2 billion, compared to the R5.8 billion reflected in the RAMS. It is not possible to see what drives the difference between these two estimates. The strategic road networks, which are also mapped in the model, were provided by the provinces. Interestingly, there was significant variation in how provinces determined their strategic road networks. Some provinces based their strategic road networks on the supporting roads required by key sectors within the region, while others crudely relied on functional road classifications that give more weight to higher-order mobility roads. The analysis indicates that the PRMG allocations are insufficient to cover the full extent of even the strategic road networks in a number of the provinces by themselves. For example, the maintenance requirements of the strategic road network in KwaZulu-Natal is 153 per cent of the 2017/18 PRMG allocation. This indicates the extent to which provinces must supplement the PRMG allocations.

DRAFT FEBRUARY 2018 PER – Provincial Roads 24

4.4 Road Project Model The Road Project Model, shown in Figure 9, is a tool that stakeholders can use to cost road projects. The model allows users to select either the baseline unit cost data discussed earlier, their own province specific cost data, or a combination of the two. This removes potential limitations of the model by giving provinces with better developed systems and databases the option to switch data sources, whilst allowing provinces with limited resources the opportunity to still cost projects within their region using the baseline data that they can input into the database using the “Own data” sheet. Figure 9 Snapshot of the Road Project Model

The model allows users to account for the complexity of road projects by generating tailored scenario that reflects actual project elements and the conditions in which they are conducted. Users have the choice of fourteen road surfaces, five activity types, five traffic volumes, and any combination of three environmental zones. Detail is shown in Figure 10 below. Note that all cells shaded in blue either contain drop down menus with options or allow the user to enter in data.

DRAFT FEBRUARY 2018 PER – Provincial Roads 25

Figure 10 Main cost variables in the roads project model

Cost data source Baseline

Main cost variables

Surface type 14+7mm Double Seal Activity Minor rehabilitation Road width (m) 8.6m Traffic volume Low-medium Distances (km) Total 68 Standard profile 31 High moisture (>20 Thorthwaite) 2 Steep gradient (>8%) 14 Steep + high moisture (Kms) 9 Rolling terrain 4 Rolling + high moitsure (Kms) 8 The division of work between small and large-scale contractors can also be specified, along with the distance to borrow pits, the labour-intensity of construction, the presence of additional climbing or passing lanes, community agreements and work on bridges (Figure 11). Figure 11 Additional cost variables in the road projects model. Additional cost variables

Average distance to borrow pit (km) 7 Community agreements R 0 Labour-intensive construction Yes Small-scale contractor (% of work) 20% Additional climbing lanes (km) 2 Additional passing lanes (km) 2 Work related to bridges R 0

Provision has been made for users to input any other cost items that may feature, such as the reconstruction of culvert headwalls or pedestrian facilities. These would be captured under “Unique cost variables” in Figure 12. The model also allows the user to factor in volatile variables such as diesel price, bitumen and wages.

DRAFT FEBRUARY 2018 PER – Provincial Roads 26

Figure 12 Other cost variables in the roads project model Unique cost variables

User must specify R 0 User must specify R 0 User must specify R 0 User must specify R 0 User must specify R 0

Price volatile variables

Diesel price (R per litre) R 10.84 Bitumen (R per MT) R 3,500 EPWP daily wage R 75 This leads to a complete costing exercise, with all elements of a project mapped. Deviations from the expected total cost are presented in 5 per cent intervals. When sizeable deviations occur between the estimated and actual costs of a project, users will be able to refer to specific elements of the project to justify or explain this outcome. Table 4 on the following page provides an illustrative example for a 10km medium volume slurry road, which is an example of a surface often used by provinces when upgrading gravel roads. The terrain options reflect the main conditions present in South Africa, with the option available to users to combine the terrain types if necessary – for example, a steep or rolling road can be found in an area with high moisture. The costs of the different road works are presented for each of the terrain types. Routine and periodic maintenance are the same for this surface type for all terrains, as these works include standard activities such as resealing and cleaning of the road and reserve area. The additional costs for roads in areas with a high moisture content account for subsoil drains and a layer of rock fill to stabilise the road foundation. The additional costs for steep roads account for paved side drains and segmented block paving or concrete to cope with the steep grades. Table 4 Cost comparison of the different road work activities on a 10km medium volume slurry road under alternative environmental conditions, R millions

Routine Periodic Minor Major Terrain Construction maintenance maintenance rehabilitation rehabilitation Standard R52.8 R1.0 R5.2 R12.8 R23.3 High moisture R84.4 R1.0 R5.2 R15.4 R28.0 Steep R73.1 R1.0 R5.2 R16.0 R31.5 Rolling R63.0 R1.0 R5.2 R14.4 R27.4

Figure 13 below (figure number?) provides a further illustrative example. In this example, the 10km medium volume slurry road has been split into 2.5km stretches of standard, high moisture, steep, and rolling terrains. The costs for each of the 2.5km lengths is shown in the graph. In addition, a 1km climbing lane and 1km passing lane have been added to the road. It was also assumed that the province required this project to be done using labour-intensive methods, with 20 per cent of the work conducted by small-scale contractors. It is clear from the figure that this project does not include any sections of road that are simultaneously steep and in a high moisture area, or rolling and in a high moisture area. No extra costs from community agreements featured, however these could have been added if appropriate. Figure 13 also shows that there were no bespoke cost

DRAFT FEBRUARY 2018 PER – Provincial Roads 27 elements, which are project unique costs that the user has the ability to manually input, or costs related to bridges for this project. Figure 13 Example project cost profile

Total cost 5% cost deviation 10% cost deviation 20% cost deviation R 25 000 000

R 23 044 932

R 21 124 521 R 20 000 000 R 20 164 316 R 19 204 110 R 315 360 R 18 243 905 R 1 368 750 R 17 283 699 R 0 R 0 R 1 280 000

R 15 363 288 R 1 600 000 R 15 000 000 R 0 R 0

R 3 600 000

R 0 R 10 000 000 R 4 000 000

R 5 000 000 R 3 840 000

R 3 200 000

R 0 Standard High moisture Steep Steep & high Rolling Rolling & high Community Climbing Passing lanes Bespoke Bridges LIC premium Contractor moisture moisture agreement lanes features premium The model has been developed as a framework to help officials ensure the correct specification for each project, and to set a benchmark against which cost estimates can be tested, thereby realising better value-for-money. The model is not intended as an audit tool. The estimated project costs are guidelines that should be used by stakeholders to engage with consultants and to interrogate the elements and cost drivers of each project. Where cost deviations occur, the intention is that these should be flagged and adequately explained by officials rather than noted as qualified findings. Forward-looking rather than retrospective application of the model is therefore suggested. If necessary, training workshops will be held on how provinces can use this tool to input their own cost data. 4.5 Funding scenarios Provinces are expected to supplement allocations from own funds with the PRMG. Figure 14 below incorporates findings from the Network Cost Model to analyse the MTEF funding allocations for the period 2017/18–2019/20. The first trend that needs mention is that the provinces, with the exception of Gauteng, KwaZulu-Natal, Mpumalanga and Western Cape, appear to be heavily reliant on the PRMG allocations. Many provinces have allocated considerably less own funds than the PRMG to roads. It should be noted that the maintenance backlogs are only being addressed in those provinces that have made significant own revenue contributions to the road sector: Gauteng, KwaZulu-Natal and Western Cape. Provinces therefore need to step-up efforts to supplement funding in the road sector, failing which their maintenance backlogs will continue to grow as the networks further deteriorate.

DRAFT FEBRUARY 2018 PER – Provincial Roads 28

Figure 14 MTEF fund allocations, 2017/18–2019/20

25 PRMG Own funds Estimated backlog

20

15 Rbillions

10

5

0 Eastern Cape Free State Gauteng KwaZulu Natal Limpopo Mpumalanga North West Northern Cape Western Cape According to Figure 14, the growth in MTEF funding allocations is well below the rate required to address the maintenance backlog estimated by the Network Cost Model. Although the maintenance backlog estimation is a simplified target, based on VCI categories and the average cost across the different surface types, with the actual figure adjusting each year taking into account works done in the previous year and general wear-and-tear, it remains a useful guideline. Scenario analysis suggests that a 25 per cent increase in funding is required to halve the maintenance backlog over the MTEF. This 25 per cent funding increase could be almost fully achieved if all the remaining provinces (Eastern Cape, Free State, Limpopo, North West and Northern Cape) raise their own contributions to match the PRMG allocations. A 50 per cent increase in funding would see the backlog eradicated over the MTEF. A 25 per cent increase is likely the best option for two reasons: firstly, given the lack of capacity in provinces, it would be sensible to adopt a longer 5-10-year time horizon over which to address a maintenance backlog of this magnitude, and secondly, while a 25 per cent increase in funding is a tall order, it is a feasible request from provinces.

DRAFT FEBRUARY 2018 PER – Provincial Roads 29

Figure 15 MTEF funding scenarios, 2017/18–2019/20

Estimated backlog Current Funding 10% increase in funding 120 20% increase in funding 50% increase in funding

100

80

60 R R billions

40

20

0 2017/18 2018/19 2019/20

DRAFT FEBRUARY 2018 PER – Provincial Roads 30

5. Provincial Road Project and Contract Reviews

This section discusses the key issues that must be considered at each stage of a road project to ensure that that major risks in the process are managed. The objective is to recommend measures to tighten these systems and thereby improve value-for-money, bearing in mind that road projects are comprised of various elements that are potentially vulnerable to inefficiencies and/or deliberate interference. The next section discusses the process that was followed to gather project documents. The Northern Cape provided a set of project files for 11 projects that were completed between 2013/14 and 2015/16. The Western Cape provided a spreadsheet with costings and analyses of 430 projects. These projects occurred between 2011 and 2015. Gauteng provided documents related to six projects, but a full set for only one of them. The Eastern Cape provided copies of project documents for 6 projects. A member of the project team appointed to analyse the documents gave a presentation of the review of the documents but broke ties with the team soon after without providing the write up. This happened late in the project process so we drew on the knowledge of other sector experts in the team with about 70 years combined experience to share their insights into the project process. Therefore, the observations and recommendations below are based on findings discussed during the above-mentioned presentation and the discussions with the sector experts. 5.1 PER process and feedback The proposal for this PER envisaged collecting a sample of project documents from all provinces and reviewing them to identify sources of variance in unit costs of road projects across the country. As per the routine PER project process and protocol, provinces were notified about the PER and the expectations we had of them. A sample of projects for each province, dated between 2012/13 and 2015/16, was selected from the Infrastructure Reporting Model. Members of the project team attended a provincial meeting on the PRMG, where the purpose of the project was explained. Following the meeting, we obtained contact details for contact persons in each province. E-mails were sent to those contacts, with the list of province-specific projects and a description of the documents and types of information we required. The project team set about scheduling meetings with provinces to arrange for the collection of the project documents. In brief, what occurred is as follows.  Meetings were set up with Northern Cape and Western Cape, and sufficient project information was obtained in electronic format at each meeting. Eastern Cape was visited, and the process explained. The province later made a large pack of project documents available in hard copy.  A meeting with Gauteng was set up, during which there was debate about the purpose of the project, whether the Provincial Treasury had been involved, and the fact that such detailed project information was more easily available for recent projects. Project documents were made available for 1 sample project, along with the required GIS data and RAMP. More documents were subsequently made available for 5 other projects, but not the full sample nor was the full set provided for any of these.  A meeting was set up with Limpopo, and they stated their eagerness to assist, but had no project files for the period the PER covered.  A meeting was held with Free State, during which they indicated they would collect the project files and make them available. An official confirmed that the sample was being

DRAFT FEBRUARY 2018 PER – Provincial Roads 31

collected, but later a different official called to ask what was required. The province later sent a file in a format that we could not read, despite clear guidance regarding our needs. GIS data and the RAMP were made available.  A meeting was held with North West, who assured they would assist. They followed up by sending payment certificates for the projects, which were not helpful.  Officials from KwaZulu-Natal were eager to assist, and made network maps and RAMP documents available for collection. No project data was ever supplied, however, despite numerous attempts.  We were unable to establish contact with Mpumalanga. As a result of these difficulties, the project steering committee directed us to use the data collected from the cooperative provinces and not waste time trying to collect data from provinces that were not being cooperative. The lack of cooperation from the majority of provinces is troubling, and we can only speculate as to why they were so unwilling to make project documents for past projects available. All the projects in the samples were completed projects, and therefore there was no sensitive project information such as unit prices or total costs that could be used to affect upcoming tender processes that provinces could have a motive to hide. Of more serious concern is that only two provinces, Northern Cape and Western Cape, could provide the relevant project information electronically and reasonably easily. The two provinces have very different filing and data management systems, but all the documentation could be found electronically and made available. 5.2 Road project process Returning to the focus of this section, Error! Reference source not found. below shows the sequence of events in the road project process. This is a simplification, but does capture the key events relevant to the discussion below. Figure 16: Key events in the road project process

DRAFT FEBRUARY 2018 PER – Provincial Roads 32

Before we look at each of these key events in detail, here is a summary of some facets of the project process:  Monitoring road condition and data collection: provinces must collect data on the visual condition of their roads networks and traffic volumes.  Pavement management system: provinces have RAMS, also referred to as Pavement Management Systems. These systems use data about the condition of the road networks and parameters to identify areas of the network where works are required.  Project identification: projects are selected through a combination of considering the outputs of analysis done by the RAMS or pavement management systems and application of experience and expert knowledge about the road network. This process should include an inspection of the site.  Drawing up the bill of quantities: once the scope of the project is identified, the Schedule of Quantities for the project can be drawn up. This is a list of all in the inputs required for the project.  Tender process: tender documents are then prepared, and the process followed to invite bidders.  Contract selection: during contract selection, the bids are evaluated. The short-listed bidders should be evaluated and thoroughly assessed before the contract is awarded.  Project execution: the project is implemented with oversight from the province and an independent engineer. The paper work created during the project cycle is substantial. For example a project in the Northern Cape for a rehabilitation project in 2013 with a contract price of R58 million had the following paperwork: contract (107 pages), appointment letter, completion certificate, construction report (98 pages), design parameter report (12 pages) and design standard certificate (5 pages), a bill of quantities (91 pages), four documents showing historical experience and qualifications of the bidder totalling about 350 pages, various documents motivating variation orders (10 pages in total), payment certificates (37 pages). Some information is repeated (e.g. the bill of quantities appears in the contract as well as the contract). 5.3 Monitoring road condition and data collection Correctly identifying which sections of the network to work on first requires quality and up-to-date condition data on the whole network. It is worth noting that the PRMG conditions provide clear guidance on the standards and frequency for collecting data about the road network. Collecting condition data on road networks is a technical activity that must be managed by experienced and skilled staff and contractors. Most provinces have pavement management systems or RAMS that use condition data to identify projects. These systems are capable of simulating scenarios using detailed data on the condition of the network, traffic volumes, weather conditions, etc. The accuracy and reliability of these simulations depend on the quality of the data in the systems and the underlying algorithms. Regardless of the quality and currency of the data, RAMS should be used only to guide the decision- making process. The simulations created by these systems must be interpreted by experienced engineers who are familiar with the needs of the road network. 5.4 Project identification and naming Provinces need to carefully check the set of recommended projects identified by their RAMS. While these systems should be able to prioritise the correct projects and formats, there are instances of

DRAFT FEBRUARY 2018 PER – Provincial Roads 33 misalignment – for example, roads requiring rehabilitation are sometimes listed for reseals. Much of the information input into the RAMS is supplied by inexperienced engineers, whose assessment of roads’ conditions through routine drive-bys for the VCI was imprecise. Moreover, visual condition assessments are unable to detect the condition of the sub-structure which has a significant bearing on the type and scope of work required. These systems must be seen as providing only guidance, however, provinces appear too reliant on the activity schedules that the RAMS generate. As part of the remedy for this, senior officials within the departments should be drawn on to ensure that remedial actions specified in the network level assessment are appropriate when defining actions on a project level. Moreover, senior officials should ensure that the VCI of their roads is properly disaggregated (there are many instances where single roads with varied VCIs along the route are inaccurately aggregated to a single level). 5.5 Appointment of consultants Provinces use different appointment systems. Some provinces, e.g. Western Cape, work off a roster system. system where consultants are invited by the province to register. During registration, consultants provide detailed information about the company structure, company experience, staff numbers, staff specialist skills and experience. A profile for each consultant is constructed from this data. The system matches consultants to projects they are suited for allowing provinces to invite the correct companies to bid for projects. This system levels the playing ground for small and large firms. The alternative is the tender-based system, which is less rigorous. Provinces who use this system select the consultant based on a once-off assessment of their technical capabilities and their price point. This approach necessitates rigorous background checks by provinces, especially for emerging contractors. The tender system provides better opportunities to less-experienced consultants, which is important in the South African context. Provinces must be aware that less experienced engineers may not engage critically with the schedules and projects generated by the RAMS, pick-up on any mistakes made by the consultants and have the confidence to take responsibility for challenging client (provinces). Strategic involvement of experienced consultants creates a crucial check-and-balance, as their experience is often crucial to flagging incorrect projects or work types. Reduced institutional knowledge has led to increased ‘box ticking’ in the technical adjudication of tenders. Because of the lack of in-depth analysis of a tenderer’s competency, more tenderers regularly meet the competency requirements and the only remaining criteria is cost. As such, when inexperienced consultants are appointed departments should quantify the risk this may pose to the success of the project and manage these accordingly. 5.6 Tender rules and processes The currently prescribed standard is that tenders are open for four weeks after advertisement. Experience has shown, however, that if contractors are given slightly more time – say six weeks – the tenders received by provinces tend to be more balanced and realistic. This outcome is due to the high volume of information and level of detail included in the project documents, sometimes exceeding 200 pages in the reviewed sample. If consultants are not given sufficient time to interrogate tenders, they are unable to evaluate them fully, and assume – sometimes incorrectly – that the province’s information is accurate. Additionally, in the rush to complete their applications, contractors build in price buffers to account for any uncertainties they may have about the accuracy of their price estimate. Similarly, consultants do not have time to research and consider innovative approaches that could result in better project designs.

DRAFT FEBRUARY 2018 PER – Provincial Roads 34

Provinces can accommodate longer tender periods by readying themselves in the off-seasons (for example, when it is too cold to seal roads) to release adverts sooner that call for work to begin on the dates that would ordinarily have been scheduled. This would allow the same volume of road works to be achieved despite the extended tender periods. Proforma tender documents serve as a template that covers all technical issues that need to be included and considered in a tender. This ensures that provinces cover all bases, and that bidding contractors have all the technical information required. All provinces should have a detailed proforma tender document, and ideally this should be standardised across provinces to ensure that contractors (especially smaller companies) do not waste time wading through varying tender rules before completing their bids. For example, in some provinces the tender rules allow contractors to submit alternative designs for consideration, while in others any alternative designs are only considered if the standard bid submitted by the contractor was the cheapest. SANRAL’s tender documents may be a useful starting point, as they are entirely fit-for-purpose. SANRAL, through COLTO, issued standard specifications for the performance of roads projects. These cover, in great detail, all the specifications that may be relevant for a roads project. For instance, the size of stone that should be used can be described using a known specification. When the schedule of quantities is drawn up, the inputs can be described according to the standard specifications. However, some provinces use project specifications that differ from the standard specifications. This is not irrational, as the standard specifications are slightly out of date. However, this makes it difficult for contractors, who must match the two documents during the tender phase. This also often creates inconsistencies that may need to be addressed mid-project. Short-listed bids must be put through a risk assessment and the balance of the cost structure in the bid assessed. It is quite possible that cheaper bids sometimes pose a greater risk to the province. For instance, the cheapest bid may have the highest monthly costs, which is hidden in the bottom line as the project plan submitted is shorter in duration than those submitted by other bidders. However, if the project faces delays and the project is extended (as was often the case for the reviewed projects), these monthly costs would cause this project to become more expensive than the others. Similarly, the tender evaluators should stress-test price-sensitive items on the Schedule of Quantities. Potential risk events might include seasonal availability of bitumen and exchange rate led fluctuations in the fuel price. If tenders do not withstand these potential events, then this risk needs to be considered by the evaluation team. This practice is sensitivity analyses that banks run on their clients’ positions. The Infrastructure Alignment Model, issued by the National Treasury as part of the Infrastructure Delivery Management Programme, may be useful in addressing some of these challenges. 5.7 Project design and specifications All departments allow contractors to claim if the quantities in the province’s schedule are amended by more than a set percentage. Because the National Treasury has strict rules about the level of cost variation from the tender price, contractors sometimes cost in contingencies equal to this set percentage to account for the risk of an incorrectly specified project. This has a cost consequence for the province. Once projects are awarded, provinces are unable to go back or amend their Schedule of Quantities, even if the project is wrongly specified. In more than half of the instances, consultants (who have professional indemnity) seem to have got the specifications wrong. While some projects items are relatively easy to specify, such as the required seal and the layer width and depth, other activities, such as materials and earthworks, are more difficult due to their respective reliance on the subbase structure and quality, and the volume of material of recyclable quality.

DRAFT FEBRUARY 2018 PER – Provincial Roads 35

The consequences of under- or over-estimation of the Schedule of Quantities can be significant, and astute contractors can profit from the consultant’s mistakes. For example, if the Schedule of Quantities drawn up for the tender documents over-estimates the volume of a specific input, contractors can enter in the correct volume and increase the unit price and still arrive at the expected total price for that activity. When the project is implemented and the lower volume expected by the bidder is realised, the province must honour the inflated price stated in the submitted Schedule of Quantities. Therefore, every purchase creates profit for the contractor, which the province has no power to curtail as the Schedule of Quantities has been signed off. There is the opportunity for consultants to intentionally make such specification errors that are difficult to pick up, and then alert select contractors to the issue for profit gain purposes. There are also instances where works are altogether not specified. When this happens, these items are not priced into the project cost. The responsibility to account for these additional costs rests with the provinces, not the consultant, as they weren’t recorded as required in the Schedule of Quantities. This issue is a common cause for provinces to overspend on contracts. A correctly-specified Schedule of Quantities is most likely to be an extensive document many pages in length. The “correct” length is, of course, project dependent. Some of the documents reviewed, however, appeared light on information and necessary details. This creates uncertainty amongst contractors, not to mention room for potential manipulation and errors. In addition, some of the information that should have been captured in the project specification sections was included in the site information section. This is problematic, as the site information section is only intended for guidance purposes, while the project specification section is what contractors are required to comply with. Mixing the two up creates confusion as to what contractors can be held liable for or not, and leads to conflict and/or cost overruns for provinces. Largely missing from the reviewed project documents was the element of road safety. In addition to the significant cost burden road accidents on provincial roads place on the Road Accident Fund, the economy and households, road deaths and injuries have now also been included as part of the Sustainable Development Goals (SDG). Improving the world’s roads to a 3-star7 or better rating is a cost-effective contribution to efforts to meet this SDG (5). The safety of a road is improved by providing proven safety features like pedestrian crossings, safe intersection layouts, safety barriers and road markings. These features are relatively inexpensive, and should be properly planned and installed during road projects and as part of maintenance programmes. However, these road safety elements were either missing from or unclearly specified in most of the projects reviewed. 5.8 Project work It is crucial that the process from bid evaluation to project implementation is not subject to any significant delays. Once projects have been delayed for a significant period, which they often are, the nature of the required remedial actions has sometimes changed. For example, if the original project was for routine maintenance but the project is delayed by a season, then the road may go past the point of reseal and require more extensive works. Engineers are contracted by the province to be an independent agent on-site to advise on and inspect the quality of the work done by contractors. However, the standard of oversight and the decisions made by these engineers are sometimes questionable. In response to an increasing number of issues related to project engineers, SANRAL have changed their rules so that engineers can only make recommendations and not binding demands. This insulates SANRAL from being bound to the financial decisions made by the project engineer. Provinces, on the other hand, have maintained authority with the project engineers, so they are accountable for their decisions. This is appropriate, given that provinces employ these engineers as skilled professionals who provide an

7 A star rating objectively measures the risk of death and serious injury on roads for pedestrians, cyclists, motorcyclists and vehicle occupants.

DRAFT FEBRUARY 2018 PER – Provincial Roads 36 independent assessment of issues that may arise. It does, however, create legal and financial risks for the provinces. Cooperation between different authorities associated with a road work can create significant cost overruns for provinces. For example, Eskom requires that only their contractors work on their infrastructure. When power lines are within a road reserve and interfere with a road project, provinces do not have a mechanism by which to procure a service provider that can move the power lines. Regardless of their own forward planning, the provinces are subject to external administrative and procurement processes. The processes of these government agencies often cause delays, which lead to significant cost overruns for provincial road departments. 5.9 Reflecting on the project process

5.9.1 A mental map of the project process Figure 17 below depicts a mental map that summarises the above discussion. Black blocks are the key components of the project process, and the flow should be read starting at the red arrow and moving clockwise through the black blocks. Figure 17 Mental map of the project process

The blocks shaded in grey highlight the activities, tools or rules that will reduce key risks.

5.9.2 Skills and experience is crucial The importance of skills and experience in each phase of the project process discussed above cannot be over-emphasised. Provincial roads departments should have a large cohort of engineers capable of ensuring quality control during every step of the project cycle. This has budgetary implications for the province in terms of compensation budgets. However, there are significant savings to be realised from the efficiencies this capacity brings about, by ensuring tender requirements and project designs are correctly specified, and effective management controls are put in place during project implementation. These savings will exceed the cost of compensation significantly. Provinces have the responsibility to maintain the knowledge and skills base required

DRAFT FEBRUARY 2018 PER – Provincial Roads 37 to make informed decisions about road network investment decisions. This is crucial to ensuring the provinces’ best interests are protected. Provinces should not be looking to the private sector for information or knowledge, which currently occurs in many provinces. Unfortunately, only a few provinces have the capacity they should. Multiple factors affect the ability of provinces to maintain a high level of internal capacity, including: migration of staff from provinces to the better-paying private sector and municipalities; early retirement packages taken by older experienced engineers – this has eroded the knowledge bases of departments; and focussed employment agendas that were implemented without recognising the time it takes to transfer knowledge from existing employees, or the time it takes to develop the expertise to make informed decisions at each stage of the project cycle. Human resource policies need to focus on creating an environment in which a pipeline of young engineers are being recruited straight from university. Western Cape, for example, provides the Masakh’iSizwe Bursaries which feed into a graduate development programme in the department. Gauteng also offers internship programmes. But budget constraints make this difficult, with no other examples of skills development programmes found. It was suggested by an engineer working in the sector that a roads department the size of KwaZulu-Natal should be sponsoring at least six engineering bursaries a year to build and maintain the level of expertise required. A high level of competence in a department ensures that correct processes and procedures are adhered to. This is the foundation on which value-for-money can be achieved. In addition, clear and transparent processes reduce the scope and opportunities for corrupt practices.

5.9.3 Activities that can be checked during oversight Based on the discussion above, the following are specific actions that could be assessed as part of the oversight mechanism on provincial roads projects. The oversight could be performed either with the provincial roads department or by the provincial department. Alternatively, these could be part of a checklist that is audited.  If the province uses a tender system, what background checks were run on the bidders? How many references were called, and how did the department ensure the references were credible? How many projects had the contractor successfully completed that were at least 80 per cent (or other per cent) of the value of this particular project?  Was a site inspection performed that was signed off by an engineer before the Schedule of quantities is drawn up? Was the schedule of quantities drawn up in accordance with standard specifications? If the answer is no, this needs to be justified.  Is the province using a proforma tender document that meets the standards of the National DoT and SANRAL tender forms?  Are the tender documents using standard specifications as per the COLTO standards?  Did the province evaluate all bids before awarding the contracts for balance? Against what standard was the Schedule of Quantities compared? Over what time frame are the sensitivity analyses of the projects applied to? Why are those time frames chosen and not longer time frames? What price variations between projects did the risk assessments reveal? Is the selected bid still the lowest cost bid if the project is delayed and takes 6, 12, 18 or 24 months to complete?

DRAFT FEBRUARY 2018 PER – Provincial Roads 38

6. Conclusions

6.1 The Road Network Model and Road Project Model The primary outputs of this PER are the Road Network Model and the Road Project Model. Both of these tools can be used by both national and provincial government, although the network model is of greater use to national role players, while the project model is better suited to the provinces’ needs. These are both tools that can help inform decision-making processes. The network model can help in the planning of PRMG and other allocations to the provincial roads sector. Provinces can use the project model to evaluate existing costing of individual projects and estimate the likely cost of future projects. It must be seen as a tool that can provide guidance. These models use the best data that was available in 2018, but this does not mean the cost estimates generated by them are sufficiently accurate for the tool to aid in the auditing of project costings. It is important to bear this in mind when using the tool to evaluate project costs. The tools have been designed to be user-friendly. However, the project team is willing to assist National Treasury and provinces by providing training on how to use the tool and how to prepare data for input into the tool. 6.2 Addressing funding shortfalls in the roads sector The discussion on the models in Section 4 shows the extent of the shortfall in provincial road funding. According to our estimates, only Gauteng, KwaZulu-Natal and Western Cape are allocating sufficient funds to address their maintenance backlogs. As the condition of all roads on a road networks are constantly degrading, this means the maintenance backlog will continue to increase and, at some point, exponentially so, in the other provinces. Our calculations showed that a 25 per cent increase in funding over the 2018 MTEF will eradicate half the backlog, and a 50 per cent increase would eradicate it within the MTEF. The expenditure analysis shows that provinces spend about 4 per cent of their total provincial budgets on roads. Achieving the 50 per cent increase requires increasing the 4 per cent to 6 per cent. This may suggest that small cuts to other budgets may yield sufficient funds for the roads sector, however, it is acknowledged that all sectors are under extreme pressure, and therefore these funds will not be easy to find. This emphasises the need for provinces to maximise their spending on their road network and achieve maximum efficiencies. Achieving maximum efficiencies from a roads budget is difficult to define succinctly. It requires, at a minimum, good data and knowledge regarding the condition of the whole network, and experienced and skilled staff making decisions and overseeing work at every stage of the project cycle. There are reasons to be concerned about the skills levels in the provincial departments, and more needs to be done to attract and retain talent within government departments. No comfort should be taken in the skills available in the private sector; provinces must be able to make their own judgments when selecting projects and defining their scope. Provinces can also address the funding shortfall by transferring an expanded number of roads of an economic significance to SANRAL. SANRAL would be required to maintain these transferred roads with existing funds in year 1, after which they can apply for additional funding from national government or for the route to be tolled. These roads also increase the value of SANRAL’s balance sheet. On balance provinces will gain financially, but there seems to be resistance to this idea from the provinces, and the reasons are not explained coherently. 6.3 Gaining more from the expenditure reporting systems A key aspect of PERs is to interrogate expenditure data reported by provinces. It was recognised before the start of this project that analyses of the BAS data would not yield the cost information required to complete the unit cost analyses that were required to develop the cost models produced

DRAFT FEBRUARY 2018 PER – Provincial Roads 39 in this project. Expenditure classifications in BAS must strike a balance between providing useful information and creating reporting burdens. However, it was evident from the analysis of the BAS data that provinces use some of the segments in different ways. In some segments, there is a high level of consistency, but, for example, in the Asset and Project segment, there were notable variations in the way it is used. It is possible to produce a significant increase in the usefulness of data recorded in BAS by making some small changes. These include adding a few reporting requirements in these segments and creating rules that ensure all the provinces use each level in the segment in the same way. National DoT, National Treasury and the provincial roads departments should work together to make these small changes so that expenditure information reported in BAS can enable better tracking of project expenditures. Suggestions on how to do this are contained in the expenditure analysis. 6.4 Minimising risks in the project cycle The discussion on the project documents identified the sources of risks in the project cycle. These are best be managed by maintaining a workforce in provincial roads departments that is skilled, experienced that conducts its work with a high degree of professionalism. Key risks identified are:  Projects are over or under specified. Road pavement management system incorrectly specifies projects, which is not corrected before the project design and schedule of quantities is finalised. Addressing this requires skilled and experienced personnel that can assess whether the specifications of the project are fit for purpose.  The selected bid lacks balance and poses a substantial financial risk to the province. Addressing this also requires skilled and personnel experience who are able to identify possible sources of cost overruns, evaluate and quantify the risks they pose to the province and address them accordingly.  Tenders are awarded to companies that lack adequate experience and skills. This can be addressed through rigorous checks on the bidders. Provinces can save time during the tender process by setting up a roster system. In addition to skills and experience, the use of uniform tender documents and standardised specifications reduce the risks of errors. Specific issues about various stages in the project cycle that could be explored by people who are not experts in the road sector were proposed. These questions are intended to be a starting point for discussions aimed to flesh out these risks and eliminate bad behaviours. The NDoT and National Treasury should work together to incorporate analyses that look at these and similar issues into the project reporting process.

DRAFT FEBRUARY 2018 PER – Provincial Roads 40

References

iRAP. 2017. Save lives with three star or better roads. Kannemeyer, L. 2014. South Africa’s Current Road Backlogs. The South African National Roads Agency: Pretoria. Prosdocimi, D., Ross, D. & Townshend, M. 2018. “Identification and Optimisation of South Africa’s Basic Access Road Network.” 20th International Conference on Geospatial Analysis, Modelling, and Simulation, 2018. Ross, D. & Townshend, M. 2017. “Cost-effective provision of low-volume roads in South Africa.” International Road Federation 18th World Road Meeting, 2017. Ross, D. & Townshend, M. 2018. “Economic classification of the South African road network.” 37th Southern African Transport Conference, 2018.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 41

7. Annexure 1 - Working Paper 1: COST-EFFECTIVE PROVISION OF LOW-VOLUME ROADS IN SOUTH AFRICA

DON ROSS MATTHEW TOWNSHEND

INTRODUCTION This paper presents a South African specific life cycle cost analysis (LCCA) of alternative unsealed and sealed surfaces for low-volume roads, defined here as carrying 75 to 220 vehicles per day, to promote their cost-effective provision by the 278 municipal and 9 provincial road authorities. The LCCA explores the trade-offs between the investment, maintenance, rehabilitation, and road user costs of gravel, sand seal, slurry seal, single chip seal, cape seal, and ultra-thin reinforced concrete pavement (UTRCP) roads. Stress tests are performed according to local variations in the proximity of natural resources to roads, the price inflation of inputs, climate, topography, the cost of labour, and road user costs to ensure this study is robust and has country-wide application. The results support a policy to seal low-volume gravel roads at a rate possible within budget limitations. Roughly 75 per cent, or 459 957 kilometres (km), of South Africa’s proclaimed road network is gravel, with an additional 131 919 km of un-proclaimed gravel roads (1). Most of this unpaved road network is classified as Class 4 rural collectors, which provide a road user access function. The remainder consists of Class 3 Provincial Trunk and Main Roads, which provide a road user mobility function. These lower order classifications translate into low traffic volumes on the gravelled networks. For example, 93 per cent of gravel roads in the Western Cape carry less than 250 vehicles per day (2). The ownership of the proclaimed gravel road network is almost evenly split between provincial and municipal road authorities. While there is limited data on the condition of municipal gravel roads, 67 per cent of the gravel roads under provincial management were in poor or very poor condition in 2013 (1). Also concerning to authorities is the rate at which the gravel road networks appear to be deteriorating, with only about 50 per cent of the sampled provincial gravel roads having been in a poor to very poor condition in 2009 (3). The South African National Roads Agency (SANRAL) estimated in 2016 that it would cost approximately R36 billion to re-gravel all unsealed provincial and municipal roads in poor to very poor condition, which is 132 per cent of the combined national, provincial, and municipal road budgets for the 2017/18 fiscal year (1,4). Despite the apparent fiscal constraints, the poor and worsening condition of the network, and low traffic volumes, municipal and provincial road authorities cannot wholly abandon this gravel road network as they are constitutionally mandated to maintain roads if they provide citizens with the only feasible means to access basic education or healthcare facilities (5). Additional sections of the low-volume gravel road network also warrant maintenance attention due to their contribution to economic activity. Given that authorities must therefore accommodate at least a portion of the deteriorated low-volume gravel road network within their available budget envelopes, it is sensible to question prior to an extensive rehabilitation exercise whether gravel, as the current default option, is the most cost-effective surface for these roads given concerns about the sustainability of the replacement rate of gravel, the relatively high road user costs, and the relevance of a capital- intensive road surface in the context of consistently high levels of unemployment. This paper applies a deterministic LCCA framework to answer this question. As advocated by Walls and Smith (6), the following procedural steps are taken in the subsequent sections of the paper to

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 42 effectively conduct the LCCA: establish common and technically appropriate surface alternatives for the analysis period; determine performance periods, activity timing, and activity costs; estimate and stress test surface costs based on realistic scenarios; develop expenditure stream diagrams; compute NPV scenarios; analyse the results; and finally evaluate surfacing strategies. THEORETICAL FRAMEWORK LCCA is an analytical technique that uses initial and discounted future costs to evaluate the overall long-term economic efficiency of competing alternative investment options (6). In the context of this paper, LCCA compares the whole-life cost of alternative road surfaces to identify the lowest-cost option that satisfies the sought performance objective. This analysis helps inform investment decisions and has thus been endorsed by several organisations, including the United States Federal Highway Administration (FHA), and notably applied by Demos (7) for the Colorado Department of Transportation (DoT), Crovetti and Owusu-Ababio (8) for the Wisconsin DoT, Lamptey et al. (9) for the Indiana DoT, and Rangaraju et al. (10) for the South Carolina DoT. To compare alternative road surfaces, LCCA uses an appropriate discount rate to convert all costs that occur throughout the life-cycle of each option to their Net Present Value (NPV). The benefits of providing and maintaining a standard pre-established road condition, along with uniform agency costs across surface alternatives - such as planning, design, and administration - are excluded from the LCCA as they are consistent across all surfaces (11). Road user costs, which include vehicle operating costs and accident costs, are addressed separately through stress tests, so the baseline model is focused on minimising road agency costs. The life-cycle costs under review thus include all differential planning, design, construction, periodic and routine maintenance, rehabilitation, and salvage costs associated with each surface option. The salvage cost includes the residual value and serviceable life of a pavement at the end of the analysis period, and is calculated based on the recyclable value of the pavement if it has reached the end of its serviceable life or the remaining life as a prorated share of the last major rehabilitation cost (12). The salvage cost, which is positive, is netted from the costs to arrive at the total cost of each alternative surface design. The NPV is calculated as follows:

−푥푖 −푧 푁푃푉 = 퐶 + ∑ 푀푖 (1 + 푟) − 푆(1 + 푟) 푖 where NPV is the present value of all costs, C is the present cost of the initial construction, Mi is the cost of the ith maintenance or rehabilitation measure, r is the real discount rate, xi is the number of years from the present to the ith maintenance or rehabilitation measure, z is the analysis period, and S is the salvage value of road surface at the end of the analysis period. Inputs for the various road surface cost variables can be generated via deterministic or probabilistic approaches. A probabilistic approach accounts for the risk of variation within the individual cost assumptions, projections, and estimates by using Monte Carlo Simulation to generate multiple outcome scenarios based on random samples from the cost inputs consistent with their defined empirical distributions (6). These outcome scenarios define an overall composite NPV probability distribution for each road surface, showing the full range of possible outcomes for each variable and the likelihood with which a particular outcome will occur. While the probabilistic approach is advocated due to the natural stochastic characteristics of factors affecting road performance, its application is reliant on large volumes of data which are unavailable for South Africa. This paper therefore adopts a deterministic approach, in which fixed, discrete costs are applied for each of the road surface variables based on evidence or professional judgement of what value is most likely to occur (13). These fixed costs are collectively used to estimate the life-cycle cost for the design alternatives. Sensitivity analyses are then conducted on a selected set of the assumptions made for major cost variables to account for uncertainty of the outcomes. Although a deterministic approach precludes simultaneous variations in multiple inputs and simplifies the

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 43 degree of uncertainty associated with life-cycle cost estimates, it is the appropriate choice in the context of data constraints and as such was applied in two-thirds of the transport specific LCCA studies reviewed by Ala-Risku (14). SPECIFICATIONS AND COST DATA FOR LOW-VOLUME ROADS This study is based on the Class 3 and Class 4 road sections from the Western Cape Government’s Geometric Manual (15). Class 3 roads are the lowest class of sealed roads and are designed for annual average daily traffic (AADT) of 200 to 400 vehicles. This road class is comprised of two 3.4 metre (m) surfaced lanes, and two 0.9 m surfaced shoulders and 0.6 m roundings constructed to the same standard as the lanes. Class 4 roads have the same cross section except they are unsurfaced. The cost estimates presented in Table 1 reflect these cross-section profiles. While it is common for LCCA studies to altogether ignore the construction and maintenance works related to the verge given that many activities are common across the surface alternatives (10), this analysis accounts for the extra brush and weed control required for sealed roads to improve sight distance given higher vehicle speeds. Road markings on sealed roads are also considered. The FHA (6) extends road design to include all pavement layers, which is appropriate for several reasons: the decision to seal a gravel road requires an upgrade of the pavement structure; these additional layers affect rehabilitation costs; and the severe deterioration of South Africa’s gravel road network often necessitates rehabilitation activities. The TRH 4 Manual (16) specifies that roads carrying between 75 and 220 mostly light vehicles per day should have a design bearing capacity of 0.1 to 0.3 million equivalent standard axles (ES0.3) per lane. The untreated ES0.3 pavement cross-section is divided into 5 elements: subgrade (150mm of G10 gravel/soil); selected (150mm of G9 gravel/soil); subbase (125mm of G6 natural gravel); base (125mm of G4 crushed or natural gravel); and a wearing course or surface. Except for the base, which is not necessary for gravel roads, the pavement structure should be prepared in the same way for low-volume gravel and sealed roads if identical traffic volumes are assumed (17). This assumption does, however, ignore the fact that a sealed road may attract diverted traffic and induce additional investment. If the pavement structure is prepared in the same way, any use of naturally occurring soils and gravels (which are generally 25 per cent cheaper than crushed stone), compaction, or cement treatment are constant across the surface alternatives. Table 1 captures the empirical data supplied by three experienced pavement engineers for the low- volume road surfaces considered in this study. The data was also cross-checked by representatives from SANRAL and the Southern African Bitumen Association (SABITA) to ensure its accuracy. Given the similarity in their characteristics and the results, sand seals are illustrative of slurry seals throughout this paper. The data is recorded per activity and disaggregated by the key inputs. The estimates presented in Table 1 assume that the road has a flat gradient, is in a moderate climate zone, and an average of 7 km from the closest borrow pit. In the absence of a geo-referenced registry of borrow pits, Ross and Field (18) assumed an average distance of 7 km between borrow pits based on reference to a Namibian study and consultations with three experienced South African pavement engineers. The applied maintenance strategies prioritise more frequent but less intensive maintenance interventions. This strategy responds to the nonlinear deterioration of roads by timeously addressing skid resistance and riding quality issues rather than waiting until the road condition is so impaired that it requires expensive rehabilitation actions (19). While this empirical data shows that gravel roads are relatively more capital-intensive than the alternative surfaces, it also suggests that local contractors have not managed to optimise the labour-intensity of construction and maintenance works on sealed roads as touted by, amongst others, SADC (20). This signals that future data might weight the labour component for sealed roads more heavily, especially given government’s attempts to promote community works programmes and small-scale contractor development. Projects using UTRCP have managed to absorb more than double the relative volume of unskilled workers as gravel roads.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 44

TABLE 1 Pavement work schedules and costs, 2017 (Rands per km)

14 mm cape seal 14 mm + 7 mm Gravel Sand seal UTRCP + 1 slurry double seal

Serviceable life 8 years 20 years 20 years 20 years 30 years

Construction

Total cost R 3 000 000 R 3 500 000 R 4 000 000 R 4 250 000 R 6 500 000

Haulage R 150 000 R 374 500 R 376 000 R 374 000 R 344 500

Unskilled labour R 180 000 R 280 000 R 360 000 R 340 000 R 780 000

Bitumen NA R 71 887 R 109 720 R 135 635 NA

Routine maintenance: Minor repairs and clean-up operations, including grading and blading

Frequency 4 per year 4 per year 4 per year 4 per year 4 per year

Total cost R 100 000 R 100 000 R 100 000 R 100 000 R 50 000

Haulage N/A N/A N/A N/A N/A

Unskilled labour R 50 000 R 50 000 R 50 000 R 50 000 R 25 000

Bitumen NA R 5 000 R 5 000 R 5 000 NA

Periodic maintenance: Reseal of light seals

Frequency (years) NA 4;12;20;28 NA NA NA

Total cost NA R 350 000 NA NA NA

Haulage NA R 6 300 NA NA NA

Unskilled labour NA R 28 000 NA NA NA

Bitumen NA R 134 133 NA NA NA

Minor rehabilitation: Strengthening of the surface layer through re-gravelling, repair, and reseal

Frequency 4 years 8 years 10 years 10 years 20 years

Total cost R 300 000 R 850 000 R 1 100 000 R 1 150 000 R 1 500 000

Haulage R 28 200 R 34 850 R 37 400 R 37 950 R 19 500

Unskilled labour R 18 000 R 68 000 R 99 000 R 92 000 R 225 000

Bitumen NA R 134 028 R 98 410 R 168 577 NA

Major rehabilitation: Intensive re-gravelling and resealing to extend serviceable life

Frequency 8 years 20 years 20 years 20 years 30 years

Total cost R 800 000 R 1 550 000 R 1 800 000 R 1 825 000 R 4 100 000

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 45

Haulage R 84 800 R 91 450 R 93 600 R 94 900 R 57 400

Unskilled labour R 48 000 R 124 000 R 162 000 R 146 000 R 492 000

Bitumen NA R 31 000 R 40 290 R 34 675 NA

Source: Own calculations. MAIN DETERMINANTS OF LOW-VOLUME ROAD COSTS As traffic volumes on most of South Africa’s rural roads are relatively stagnant at AADT less than 200, the agency life-cycle costs of low-volume roads are predominantly driven by environmental factors, resource availability, and resource costs. The first of these factors is climate, primarily temperature and moisture. Effects of extreme road surface temperatures, greater than 60°C or below 0°C, on pavement behaviour and performance include cracking, permanent deformation, warping, curling, evaporation, weathering, speed of reactions, ageing, and drying out of materials (19). Extreme moisture, scoring higher than 20 on Thornthwaite’s Moisture Index, influences safety, drainage, erosion rates, permeability, material strength, and material selection (19). The implications for construction costs in wetter regions are generally a thicker pavement structure to improve the bearing capacity of the road, deeper side drains, and more sophisticated subsoil drains. High levels of precipitation also increase the frequency and cost of routine maintenance on gravel roads, and during large storms there may even be significant gravel loss. Similarly, road gradients greater than 6 per cent impact negatively on the rideability of gravel roads and require an increase in the frequency and cost of routine maintenance to retain the road’s original design life. The local price of bitumen increased with alarming volatility from the mid-2000s, driven largely by changes in the world cost of petroleum products and supply challenges caused by shutdowns at South African oil refineries. This price volatility has had a significant effect on road expenditure, as evidenced by the Gauteng Department of Public Transport, Roads and Works who in 2005 attributed the bulk of a 67 per cent year-on-year increase in the cost of sealing a low-volume gravel road to higher bitumen prices (18). Moreover, the supply shortages in 2013 meant that 15 per cent of local bitumen demand was covered through imports at a premium of R1500 per ton (21). Except for the price of fuel, which is similarly sensitive to world oil prices, the price of the remaining road surface materials have roughly tracked the building and construction sector price index (22). Haulage, more through the marginal increase in fuel costs than through rental or depreciation of delivery vehicles, affects the life-cycle cost of low-volume roads. Maintenance of the local gravel road network requires approximately 30 million cubic metres of aggregate material per annum, which at an average distance of 7 km between borrow pits equates to about 30 million litres of fuel per year (18). Sealed roads also incur haulage, but the cost is marginal and infrequent following the construction phase. In fact, Ross and Field (18) and Jahren et al. (11) found that sealed roads accrued lower oil-based costs than gravel alternatives once the haulage distance exceeds 11 km. While this specific distance is sensitive to shifts in oil prices and the exchange-rate, what is evident is that any change in haulage distance is likely to have a greater proportional impact on the cost- effectiveness of gravel roads. Given the onerous environmental impact assessments (EIAs) required by the Mineral and Petroleum Resources Development Act (Act 28 of 2002) to open new quarry facilities and the associated delays in the approval of mining permits, the recent trend has been towards a more uneven spread of borrow pits and thus increased haulage distances (23). The high unemployment rates across South Africa reflect an abundance of underutilised labour that government is trying to mobilise through the Expanded Public Works Programme (EPWP), which calls for the application of labour-intensive production methods wherever the accounting cost is no higher than alternative capital-intensive methods. Ross and Field (18) argue that this prescription ignores the possibility of building the national asset stock by investing in human capital. The training provided to EPWP workers on road projects improves the productivity of individual contracting firms. This skills development also translates into higher potential national productivity and should thus

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 46 be subsidised by government. The cost of labour from government’s perspective should therefore not reflect the accounting wage of R83 per day for an EPWP worker in 2017, which has been artificially inflated above the market-clearing rate by labour unions and minimum wage regulation, but the opportunity cost of a potential EPWP worker’s time, measured as the cost to the economy in terms of foregone output from moving that person from their present occupation to employment on an EPWP project. When these shadow wages are considered, the relative capital- and labour- intensity of the road surface has a significant effect on the LCCA. Finally, discounting is an influential element of the LCCA. Higher discount rates lead construction costs to dominate maintenance costs, and vice versa. Higher discount rates therefore favour gravel roads, which have relatively low construction costs but frequent and relatively high maintenance costs, over sealed alternatives. Fortunately estimates of the discount rate in South Africa are within a relatively narrow band. The National Treasury’s working rate of 9 per cent and Kuo et al.’s (24) estimation of 11 per cent are roughly in line with the World Bank’s and Asian Development Bank’s recommendation of 10 to 12 per cent for developing economies (25). A standard discount rate of 10 per cent is thus assumed throughout this paper. LCCA RESULTS The analysis period, which is the time horizon over which the pavement designs are evaluated, was set at 30 years to cover the longest design life, to incorporate at least one major rehabilitation activity per surface, and to reflect long-term cost differences associated with the alternatives (6). The tests that follow individually stress relevant cost determinants, and then introduce road user costs to guide authorities on a cost-effective surfacing policy under variations in local conditions. Figure 1 presents the life-cycle cost profile of the alternative road surfaces in constant 2017 prices. This scenario, which serves as the baseline for this study, indicates that gravel roads are the most cost-effective surface option under the simplified standard conditions: flat road, moderate climate, and an average distance of 7 km between borrow pits. While the recurrent re-gravelling lowers the opportunity cost between gravel and sealed roads over the analysis period, the initial cost savings in the construction of gravel roads is the dominant factor once discount rates have been considered.

FIGURE 1 LCCA under standard conditions The first stress test, which is not shown graphically as the results align with the baseline outcome in Figure 1, accounts for a scenario in which no significant bitumen supply is generated as a by- product of local fuel refining. While some additional bitumen storage capacity has been added following the supply crises in 2011 and 2013, the rationale for this scenario is the persistent risk of maintenance shutdowns at South African oil refineries and that growth in local oil refining volumes

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 47 will not keep pace with growth in bitumen demand. To test the cost implications of this eventuality, price inflation was controlled for by applying an annual inflation rate of 4.4 per cent for the building and construction sector to all inputs except bitumen (22). This inflation rate was based on the annual average between 2011 and 2017, which excludes the price volatility introduced by the infrastructure drive in advance of the 2010 Soccer World Cup. The annual inflation rate of bitumen prices was set at 18 per cent, which captures the average between 2011 and 2014 when the recent supply shortages occurred. While gravel remains the cheapest surface option under these conditions, this marked increase in the price of bitumen did not cause a significant fluctuation in the life-cycle cost of sealed roads as bitumen is a relatively small component of the materials bills. High moisture content and high precipitation volumes have a notable effect on the relative cost- effectiveness of gravel roads. A similar effect is evident for road gradients steeper than 6 per cent, hence the results shown in Figure 2 relate to both scenarios. Both phenomena generally necessitate two additional grading events per annum for gravel roads. High precipitation volumes also inflate the construction cost of sealed roads by 5 to 7 per cent to fund upgraded drainage and pavement strengthening. Despite the increased cost to construct a sealed road, in both scenarios the cost of the additional grading events erodes the initial construction savings on gravel roads by the eighth year. Authorities whose jurisdictions experience high volumes of precipitation or include hilly and mountainous areas should therefore consider adopting a policy of sealing gravel roads.

FIGURE 2 LCCA in areas with high precipitation volumes or steep road gradients Haulage is a significant cost component of initial construction outlays for both gravel and sealed roads. While haulage falls away somewhat as a proportion of the rehabilitation costs for sealed roads, it remains a major cost driver throughout the life cycle of a gravelled road. Figure 3 shows that at the current diesel price of R10.84 per litre, a 15 km increase in the average distance between borrow pits would level the life-cycle costs of gravel and sand, slurry, and 14mm cape seal roads. It is therefore evident that sealed roads will become increasingly preferable should the EIA related pressures surrounding the opening of new borrow pits not subside.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 48

FIGURE 3 LCCA with a 15km increase in the average distance between borrow pits The test in Figure 4 substitutes the EPWP wage for the shadow price of unskilled labour in South Africa. In light of the preliminary nature of estimates of the regional shadow price of labour, we here opt to set the shadow price of unskilled labour at 50 per cent of the EPWP wage for illustrative purposes. The fact that contractors have not maximised the labour-intensity of road works skews the analysis in favour of gravel roads, which in theory should be more capital-intensive than sealed alternatives and therefore a less attractive surfacing option given high unemployment rates and the low levels of informal sector productivity. The test therefore also artificially enhanced the labour- intensity of road works to demonstrate that a four-fold increase in the labour-intensity of works on sealed roads is required to roughly equalise the economic cost of providing gravel and sealed roads.

FIGURE 4 LCCA using the shadow price of labour and enhanced labour-intensity Figure 5 incorporates the combined effects of road accidents and vehicle operating costs into the LCCA, assuming a traffic volume of 75 mostly light vehicles per day and a speed limit of 80 km/h. Working from World Bank (26) statistics on surface specific road accident rates in South Africa per 100 million vehicle-km, the 821 250 vehicle-km travelled per 1 km stretch of modelled road over the 30-year analysis period would lead to approximately 1.89 accidents on gravel roads and 0.82 accidents on surfaced alternatives. For the purposes of illustrating the full potential cost of road accidents, we here assume that the one additional accident on gravel roads is fatal and apportion

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 49 the cost of this incident - valued by the Road Traffic Management Corporation in 2015 at R5.44 million (27) - over the analysis period. In addition, the roughness of gravel roads can increase vehicle operating costs through fuel and oil consumption, depreciation and interests, tyre wear, and maintenance and repairs. This test, however, is restricted to fuel consumption only as this tends to be the most significant and transparent vehicle operating cost. At an average fuel consumption of 9.89 km per litre (29), this equates to 2 768 litres of fuel per annum to cover the 27 375 vehicle-km. Studies have shown that average fuel consumption over different weather conditions and phases of the road maintenance cycle is approximately 8.8 per cent higher on gravel roads (28). At the current price of R10.84 per litre of diesel, an extra 8.8 per cent fuel consumption equates to R2 640 per annum. Despite this limited scope in vehicle operating costs, the results in Figure 5 still show a marked deterioration in the relative cost-effectiveness of gravel roads once the effects on road users are considered.

FIGURE 5 LCCA accounting for fuel consumption and road accident costs CONCLUSION Additional benefits of sealing gravel roads include: faster vehicle speeds resulting in shorter travel times; a potentially enhanced tax base stemming from increased prices of neighbouring properties; reduced dust emissions and a subsequent reduction in cases of eye and respiratory issues; and better vegetation and crop growth on adjacent land (11). These factors should improve the relative cost-effectiveness of sealed roads and therefore shorten the time period within which the upfront investment in sealing of gravel roads is recouped. Our baseline condition represents an ideally simple world for surfacing decisions. A few regions of South Africa, like the Karoo, might approximate this world. However, such regions are sparsely populated and are not the sites of many factor-sensitive surfacing decisions. Policy should thus not be driven by the ideally simple case, but by reference to tendencies in the relationship between complicating factors and relative surface construction and maintenance costs. All factors examined here individually eliminate the cost advantage accruing to gravel in the ideally simple world. There is near-unanimity among economists that South Africa's most urgent policy priority is improving the quality and extending the distribution of lower-skilled human capital that has a non- negative shadow value. Indeed, the policy priority to increase employment rates, community participation, and skills development through labour-intensive road projects is captured as part of the Provincial Roads Maintenance Grant (PRMG) outcomes (30). Such assessment should be done in terms of utility, not monetary value. However, there is no meaningful room for doubting that, in South Africa, if a policy A dominates or ties with a policy B in terms of expected monetary value, but A contributes more to the human capital stock amongst citizens with relatively low levels of

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 50 formal education, than A dominates B with respect to optimizing public utility. Our analysis indicates that choosing sealed surfaces over gravel surfaces is an "A-type" policy where the overwhelming majority of real road surfacing decisions are concerned. One of the requirements attached to the PRMG, which comprised 44.3 per cent of total provincial road expenditure in 2016/17 and as much as 77 per cent in the Northern Cape, is that the funds are allocated to rehabilitation or routine, periodic, and special maintenance (30). The requirements in fact stipulate that the PRMG cannot be used to upgrade gravel roads to surfaced roads, meaning that provinces have had to fund road upgrades from their equitable share or own revenues. While this restriction has had an obvious stifling effect on the surfacing policy adopted by provincial road authorities, a newly proposed version of the PRMG framework seeks to relax this condition and add ‘number of kilometres of roads upgraded’ as an outcome target measure. This amendment would provide necessary financial support to allow more rapid sealing of gravel roads. The surfacing policy promoted by this paper is aligned with all the general conditions required for the sustainable provision of low-volume roads (20). Over-and-above the fact that the labour- intensive construction and maintenance works on sealed roads can be decentralised to small contractors and local communities, in line with government policy, the techniques associated with these works are technologically appropriate in the South African context. Moreover, internal human resource constraints have already led some road authorities – for example the KwaZulu-Natal DoT – to promote sealed surfaces to minimise the maintenance commitment of low-volume roads through their design life. Our findings thus give support to the proposition that, in South Africa under its current economic and institutional conditions, if a road is worth maintaining at all it is worth sealing. REFERENCES 1. Kannemeyer, L. 2016. South African Road Network Condition, Needs and Funding. The South African National Roads Agency: Pretoria. 2. October, C. 2016. A decision-support process for the upgrading of gravel roads in the Western Cape Province. 80th IMESA Conference: East London. 3. Kannemeyer, L. 2014. South Africa’s Current Road Backlogs. The South African National Roads Agency: Pretoria. 4. National Treasury. 2017. Estimates of National Expenditure. National Treasury: Pretoria. 5. Ross, D., and Townshend, M. 2015. Can South African road authorities satisfy constitutionally guaranteed basic access rights without unduly sacrificing economic growth? 15th Conference on Asphalt Pavements in Southern Africa. 6. Walls, J., and Smith, M. 1998. Life-Cycle Cost Analysis in Pavement Design. Interim Technical Bulletin. Federal Highway Administration: Washington, DC. 7. Demos, G. 2006. Life Cycle Cost Analysis and Discount Rate on Pavements for the Colorado Department of Transportation. Colorado Department of Transportation: Denver, CO. 8. Crovetti, J., and Owusu-Ababio, S. 1999. Investigation of Feasible Pavement Design Alternatives for WisDOT. Wisconsin Department of Transportation: Milwaukee, WI. 9. Lamptey, G., Ahmad, M., Labi, S., and Sinha, K. 2005. Life Cycle Cost Analysis for INDOT Pavement Design Procedures. Indiana Department of Transportation: West Lafayette, IN. 10. Rangaraju, P., Amirkhania, S., and Guven, Z. 2008. Life Cycle Cost Analysis for Pavement Type Selection. South Carolina Department of Transportation: Columbia, SC.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure I: Working Paper I 51

11. Jahren, C., Smith, D., Thorius, J., Rukashazza-Mukome, M., White, D., and Johnson, G. 2005. Economics of Upgrading an Aggregate Road. Minnesota Department of Transport: St. Paul. 12. Caltrans. 2011. Using Life Cycle Cost Analysis in Highway Project Development. Caltrans: Division of Research and Innovation. 13. Rahman, S., and Vanier, D. 2004. Life cycle cost analysis as a decision support tool for managing municipal infrastructure. National Research Council of Canada. 14. Ala-Risku, E. 2008. Life cycle costing: A review of published case studies. Managerial Auditing Journal: 23:3. 15. Western Cape Provincial Administration. 2006. Gravel Roads Manual. Department of Transport and Public Works: Roads Infrastructure Branch. 16. COLTO. 1996. TRH 4: Structural Design of Flexible Pavements for Interurban and Rural Roads. Department of Transport: Pretoria. 17. Ethiopian Roads Authority. 2011. Design Manual for Low Volume Roads. Ethiopian Roads Authority: Addis Ababa. 18. Ross, D., and Field, K. 2007. South African road surfacing policy, international oil price changes, and the shadow pricing of costs and benefits. Prepared under contract to the Southern African Bitumen Association. 19. SANRAL. 2013. South African Pavement Engineering Manual. SANRAL: Pretoria. 20. Sothern African Development Community. 2003. The SADC Guideline of Low-Volume Sealed Roads. SADC: Gabarone. 21. Cokayne, R. 2013. Bitumen shortage is a threat to roadwork. Business Report. 22. Statistics South Africa. 2017. Contract Price Adjustment Provisions Work Group Indices, May 2017. Statistics South Africa: Pretoria. 23. National Treasury. 2006. Medium Term Expenditure Framework Treasury Guidelines: Preparing Budget Proposals for the 2007 MTEF. National Treasury: Pretoria. 24. Kuo, C., Jenkins, G., and Mphahlele, B. 2003. The Economic Opportunity Cost of Capital in South Africa. The South African Journal of Economics: 71:3. 25. Zhuang, J., Liang, Z., Lin, T., and De Guzman, F. 2007. Theory and Practice in the Choice of Social Discount Rate for Cost-Benefit Analysis. Asian Development Bank: Philippines. 26. World Bank. 2010. HDM-4 Road User Effects. World Bank: Washington, D.C. 27. Road Traffic Management Corporation. 2016. Cost of crashes in South Africa. Road Traffic Management Corporation: Pretoria. 28. Savenhed, H. 1986. Vehicle Fuel Consumption of Different Types of Wearing Courses. Swedish Road and Traffic Research Institute: Linkoping. 29. Federal Highway Administration. 2013. Highway Statistics 2013. Federal Highway Administration: Washington, D.C. 30. Republic of South Africa. 2017. Division of Revenue Bill. Government Gazette No. 40610.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 52

8. Annexure 2 - Working Paper 2: AN ECONOMICS-BASED ROAD CLASSIFICATION SYSTEM FOR SOUTH AFRICA

Don Ross and Matthew Townshend

1 INTRODUCTION

Ross and Townshend (2015) follow the 1996 White Paper on National Transport Policy to set out the road network’s priorities: to satisfy citizens’ constitutional right to access basic services; and to maximise potential economic growth. But current road classification systems, which are covered in Section 2, inadequately account for either function. The implication of this oversight is potentially inaccurate road prioritisation, which in the context of limited budgets would lead to inefficient public expenditure. This paper therefore follows fundamental classification criteria listed in Section 3 to develop an economics-based road classification system for the South African road network, with focus on the definition and identification methodology for each class. The final section describes how this information can be incorporated into asset management systems to improve expenditure and investment outcomes.

2 CRITIQUE OF CURRENT CLASSIFICATION SYSTEMS

Road classification systems categorise roads and order them in relation to each other. Amongst other contributions, these classification systems should improve the administration, financing, and management of road networks. Constrained budgets, which are a reality in most countries including South Africa, force authorities to choose between roads. In this context, it is imperative that the classification system identify and prioritise roads within the network that respectively facilitate access to constitutionally protected public services and maximise economic growth. This is reiterated by the National Department of Transport (NDOT) (2004), who state that road classification cannot be considered from a one-dimensional perspective that does not account for the socio-political and economic imperatives of the country. Because basic access and economic growth are the primary functions of the road network, the classification system must clearly define and individually identify these services within the classes. The South African road network is primarily classified according to administrative, geometric, and functional characteristics. Additional classification systems exist, such as naming, route numbers, and road traffic signs. Each system is designed for a unique purpose, but none are individually or cumulatively sufficient to support economic prioritisation of the network. 2.1 Naming hierarchy

The naming hierarchy classifies roads according to their names. Numbered mobility roads are the first class and are referred to as a freeway, bypass, motorway, expressway, route, highway, or arterial. The next class are numbered and named mobility roads and include avenues, roads, drives, and links. Access and activity streets are the lowest class and identified as boulevard, collector, street, lane, loop, crescent, place, way, terrace, close, grove, court, square, or mall. The main benefit of this system is public information and management of users’ expectations. From an economic prioritisation perspective, however, the classes provide little detail about necessary characteristics like traffic volume, potential cargo, and connected service centres. 2.2 Road Traffic Signs Manual classification

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 53

This system categorises roads into four classes: Class A1 dual carriageway freeways; Class A2 single carriageway freeways; Class B1 non-freeway numbered national, provincial, regional and metropolitan routes; Class B2 un-numbered surfaced routes; Class C1 low volume surfaced routes, local collector-distributor streets; Class C2 un-numbered gravel and industrial streets; and Class D local access roads with no public destination. The classes are roughly based on traffic volumes, which is a proxy indicator for economic activity. But there are several instances where roads in lower design classes should be viewed as more productive assets than roads in higher classes. For example, some numbered routes in densely populated cities carry more traffic than single and dual carriageway freeways in smaller centres. Moreover, no provision is made within these classes for basic access roads. 2.3 Route number classification

The route numbering system expresses ownership, with the prefixes N, R, and M for national, provincial, and municipal roads, respectively. Provinces further differentiate their roads by using D for district roads, T for tourist routes, and respectively adding two- and three-digit numbers to major and minor R routes (COTO, 2012). These identifiers are no longer accurate due to the transfer and reassignment of roads amongst authorities. Route numbers help guide motorists, but are inadequate for prioritisation due to the high incidence of exceptions, inconsistencies, and un- numbered routes in the system (COTO, 2012). The information is also at a high level, meaning that route numbers are not tied to the service or value provided by a road. 2.4 Administrative classification

Administrative classification categorizes roads according to the responsible authority, either national, provincial, or local government. Each authority is accountable for management of their designated road network. There is, however, no standard methodology to determine the assignment of roads. Current legislation simply defines national, provincial, and municipal roads as those proclaimed as such with no set rules for doing so (COTO, 2012). Despite the revised inception of this classification system post-1994, approximately 144 000 kms remain as unproclaimed roads (Kannemeyer, 2016). Unwillingness by authorities to assume responsibility for these orphaned roads means that potentially important basic access and mobility roads cannot be legally maintained using public funds. Moreover, the necessary information to inform onboarding and upgrade schedules for these unproclaimed roads is missing within the system. The NDOT (2006) noted that administrative classification cannot coordinate overlap of services between roads assigned to different authorities. The origins and destinations of many economic activities are on sub-national roads, making them part of strategic supply routes. But this system only classifies national roads as strategic assets. The NDOT has thus called for a new system that can identify strategic roads at all levels of administrative classification. 2.5 Geometric classification

Geometric classification draws on structural determinants such as road width, surface type, wet weather condition, gradient, load bearing, and height restrictions to class roads according to their design (Intergovernmental Committee on Surveying and Mapping, 2006: 5). Freeways, for example, are in a higher class than dual carriageways, which in turn are above undivided arterials. The system is complicated by the fact that different segments of the same road can have different structural determinants and geometric design. The system is used in South Africa to aid design processes and communication between engineers and administrators. No direct relationship, however, can be drawn between geometric road types and economic contribution. The NDOT (2006) actually state that undivided, two-lane roads can be more important for freight and passenger mobility than geometrically superior dual-carriageway multi-lane collectors.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 54

2.6 Functional classification

Functional classification uses expected traffic movement to rank roads. The system assumes that roads perform two types of functions, either providing traffic mobility or land access. Mobility roads are higher speed through-routes that link centres of economic activity, and because of this have limited access points as this slows traffic. Access roads, on the other hand, are shorter in distance and cater specifically for access to land, activities, and services. The proportion of mobility and access provided by a road determines the functional class to which it is assigned. The NDOT began with the functional road classification system in the early 2000s due to the need for a uniform classification methodology across the spheres of government. TRH 26 (COTO, 2012), which is the official functional classification manual, built on the Department’s work to generate six road classes. Three criteria were used to distinguish between the road classes. The size and strategic value of the trip generator is the first criterion. According to COTO (2012), mobility roads link large or important trip generators and rural and urban centres of development. Access roads give direct or indirect access to properties and collect and distribute traffic between those properties and mobility roads. The next criterion is the reach and connectivity of the road. While mobility roads cater for longer travel distances, access roads only facilitate shorter trips. To avoid speeding in urban areas, access roads should not exceed 1 km before connecting with a mobility road. Lastly, the travel stage is considered. Trips are undertaken in three distinct stages: local at the origin, through when away from the origin or destination, and local again at the destination. COTO (2012) state that the local portions of a trip should be on access roads, while the through portion should be on mobility roads. The NDOT (2006) conclude the Road Infrastructure Strategic Framework for South Africa by stressing the need to develop a framework to support economic growth sectors through the provision of road infrastructure. One of the reasons this system fails to fulfil this demand is the exclusion of traffic volumes. Although high order mobility roads usually carry more traffic than lower order mobility and access roads, this cannot always be assumed. For example, the NDOT (2006) compare long-distance mobility roads required for through travel in areas with low demand to access roads connected to key service centres in densely populated urban areas. Functional classification also lacks the requisite detail to identify which facilities are serviced by a road, making it impossible to systematically identify the roads connected to basic services essential to some specific population. It is erroneous to treat all Class 5 roads as equivalent or to assume that all Class 4 roads are more important than Class 5 roads, as each road provides a unique service that will affect the prioritisation exercise. As such, the NDOT (2004: 15) admit that this functional classification system was “not intended to suggest how priorities and funding allocations should be made”.

3 FUNDAMENTAL CLASSIFICATION CRITERIA

The Intergovernmental Committee on Surveying and Mapping (ICSM) (2006), which published a seminal report on road classification developments, list a set of fundamental criteria for an effective road classification. The first criterion is a small number of road classes, allowing officials to effectively manage and work with the system. The ICSM note the need for trade-offs between simplicity and accommodation of all classes. Practically relevant road classes must not be excluded in the process of rationalising the theoretically possible classes. The second criterion is that classification systems use unambiguous descriptive terminology for the road classes. The definitions should be distinct, clear, and concise to ensure simple and objective application of the classification system. Broad definitions of classes allow more scope for interpretation and thus impair consistent application of a classification system.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 55

The third criterion is that classes are scaleless and ubiquitous across the network. The entire network must be consistently classified, not sections in isolation from one another. Modifications to classes to account for regional significance detract from the countrywide effectiveness of the system. The ICSM assert that a system should only contain distinguishing variables that are applicable across the whole spectrum of roads. Unique variables, such as those specific to national roads, should therefore be avoided irrespective of their identification or quantification ease. The final criterion is hierarchical contiguity, which requires that roads of the same class connect to form a continuous network. The idea is that by creating seamless networks, the classification system would ensure that assignment of ownership would allow each responsible authority to schedule maintenance in such a way as to preserve network integrity. However, this criterion undermines the efficiencies that are obtainable from prioritising across the South African road network as a whole. It could enjoin maintaining or upgrading a tactical road A only because it connects tactical roads B and C, even in a case where B and C are relatively efficiently connected by a strategic road D. This system therefore does not aim to satisfy this criterion.

4 ECONOMICS-BASED ROAD CLASSIFICATION SYSTEM

The proposed economics-based road classification system has four classes: Basic Access Roads; Strategic Roads; Tactical Roads; and Surplus Roads. These classes are designed to identify: roads that primarily satisfy citizens’ constitutional right to access basic services; roads that maximise potential economic growth; and roads with a negative economic contribution. The prioritisation rule for these new road classes, shown in Figure 1, is based on the normative reasoning applied by Ross and Townshend (2015). Basic Access Roads are the first lexical priority, ensuring that all citizens enjoy at least the minimum level of access to constitutionally protected primary and secondary schools and healthcare facilities. The second lexical priority is contribution to economic growth. Both Strategic and Tactical Roads perform an economic growth function, but the former are prioritised due to their higher and structural traffic volumes. Surplus Roads make a negative economic contribution, meaning that their maintenance cost exceeds the value of the economic activity they facilitate. Authorities are advised to unproclaim Surplus Roads, thereby freeing the sector’s resources for more efficient allocation to higher priority roads. A road is essential for basic access only if it is the only viable means for communities to access basic service centres, and in its absence citizens fall outside prescribed norms and standards for access to these facilities. Government's constitutional obligation prevents it from being able to trade off basic access for contribution to economic growth. Thus, Basic Access Roads that make low contributions to growth must be maintained in the top priority tier nevertheless. Of course, some roads, which we call 'multi-functional', are essential for basic access and also contribute to economic growth. The more such roads there are, the lower is the medium-run fiscal drag due to the morally mandated constitutional requirement. Single-function Tactical Roads should be the final call on resources, with Surplus Roads unproclaimed.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 56

Figure 1: Prioritisation rule 4.1 Basic Access Roads

Without the option of alternative access routes to basic service centres, households are reliant on authorities to ensure Basic Access Roads are maintained. The provision of these roads is a constitutional obligation, as well as a tool to help slow rates of rural-urban migration within manageable limits. As cyclical economic fluctuations have no bearing on the need for children to attend school and people requiring medical attention to reach healthcare facilities, the demand elasticity for these roads is zero. Although this area of demand should be seasonally consistent, withstanding the effects of migration, it is possible that factors such as poor road conditions, a non- all-weather road surface, and the availability of public transport might force users to alter their travel patterns. The procedure to identify the quickest access routes, which together with traffic data also indicates when users are forced to use longer routes, is to geospatially locate all households in the region and map the roads which form the quickest route from their place of residence (or closest point to it) to the nearest public primary and secondary school and healthcare facility. Because traffic data is missing for large sections of the South African network, the potential Basic Access Road Network presented in Figure 2 was estimated using 2011 Census data at the enumeration area level, GPS coordinates for the 25 137 registered ordinary and special needs primary and secondary schools and 5 389 healthcare facilities, and the additional conditions for Basic Access Roads introduced below. Urban roads are excluded as Basic Access Roads given the availability of alternative routes to service centres in these areas. Geospatial land use data, such as that prepared by the CSIR (2018), was used to remove all urban roads from the Basic Access Road Network. The naming hierarchy is a useful, albeit less accurate, reference as certain road names, such as Avenues, Lanes, and Streets, are generally associated only with urban areas. The identification methodology also applies the gazetted access norms and standards, which stipulate that public primary and secondary schools must have a feeder zone with a radius of 5- kms (Government Gazette 33283, 2010). Because this zone reflects the acceptable walking distance to schools, it creates a buffer within which authorities are not constitutionally required to

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 57 provide access roads. No formal standard exists for healthcare facilities, which is sensible as it is unreasonable to expect people who require medical care to walk non-trivial distances. Bearing in mind the importance of direct and all-weather road access to schools, authorities should include the lowest volume of direct access routes within 5-km of all households. As mentioned, some roads will simultaneously fulfil a basic access function and contribute to economic growth. These multi-functional roads are efficient as they allow authorities to address access needs at the same time as promoting economic growth. While overlapped basic access, strategic, and tactical road functions can occur in any rural setting, most multi-functional roads are likely to be in areas with higher economic activity – in this case measured using 2010 Gross Value Added (GVA) data from the CSIR (2017). Figure 2 thus indicates the possible distribution and concentration of multi-functional versus single-function Basic Access Roads.

Figure 2: The potential Basic Access Road Network, 2018. 4.2 Strategic Roads

Strategic Roads include large-volume transport routes, which generally carry more than 10 000 vehicles per day within or between key areas and locations. These includes cities, major towns, international and local trade corridors, and high-volume freight and passenger terminals. While the concept of a Strategic Road Network is often used to refer to national roads, this network should extend to all roads that are core enablers of economic activity. Accordingly, the Road Network Study identified a 9 200 km Core Strategic Network, 9 600 km Secondary Strategic Network, and 14 000 km Primary Road Network (SANRAL, 2015). The Secondary Strategic Network provides alternative routes to the Core Strategic Network, which are required in areas where low road density makes the Core Strategic Network difficult to reach. The Primary Road Network feeds the two strategic road networks. The Core and Secondary Strategic Networks are designated as national roads, while the Primary Road Network is a combination of provincial and national roads. Importantly, user demand for these roads is inelastic. Adding this criterion to Strategic Roads

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 58 prioritises the maintenance of roads that make the highest contribution to economic growth and support structural economic activities over cyclical alternatives. It follows that urban roads, which have similarly inelastic demand, are classified as Strategic Roads. From an economic perspective, urban roads also combine to form integrated networks that support vital daily mobility and access functions; including the transportation of all goods consumed within urban boundaries and the movement of residents between their homes, public facilities, and places of work. Incorporating urban roads in the Strategic Road Network aligns maintenance schedules with population density, regardless of traffic data availability, and ensures that asset management systems adequately reflect urban sprawl pressures. Figure 3 provides an illustrative estimation of the Strategic Road Network. This example of the network combines the 32 800 km of Core Strategic, Secondary Strategic, and Primary Roads with all urban roads. The urban roads were identified by overlaying the CSIR’s 2018 geospatial land use data with the total road network. Additional provincial and municipal roads may warrant inclusion in the Strategic Road Network following more detailed economic analysis.

Figure 3: The estimated Strategic Road Network, 2018. Survey responses provided by senior provincial road officials and the provincial Road Asset Management Plans reveal that many authorities have based their strategic road networks on functional classifications. Unfortunately, classification as a higher-order Class 1 or Class 2 mobility road is insufficient to warrant inclusion within the Strategic Road Network. Firstly, it is imprecise to treat all roads within a class as equivalent as each road has a distinct demand profile and makes a unique contribution to economic growth. Secondly, some of the urban roads which form part of a holistic Strategic Road Network are classified as lower-order access roads.

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 59

4.3 Tactical Roads

Tactical Roads comprise rural and peri-urban roads that facilitate low to medium volumes of cyclical traffic. Although the value of freight transported on these roads is generally lower than Strategic Roads (there may be some exceptions for urban roads), NDOT (2006) stress that an effective Tactical Road Network is required for the sector to fulfil its mandate as service centres and the origins, trade routes, and destinations of many economic activities are located along secondary and tertiary roads that are not part of the Strategic Road Network. The next criterion is that Tactical Roads make a positive contribution to economic growth. This determination should be based on a structural growth model that comprises two terms: the expected cost of the road; and the roads contribution to economic growth. The cost term should reflect the present value of the future cost of road maintenance, bearing in mind Ross and Townshend’s (2017) conclusion that, due to the relative labour intensity of sealed road maintenance compared to gravel road maintenance, and the low shadow price of labour in parts of South Africa where there is scope for surface upgrades, if a road is worth maintaining at all it is worth sealing. 4.4 Surplus Roads

Surplus Roads neither fulfil a basic access function nor contribute positively as sealed roads to economic growth, based on the same structural growth model used to identify Tactical Roads. Because the present value of the cost of maintaining these roads exceeds the economic contribution, authorities should unproclaim Surplus Roads and reprioritise any allocated funding towards the other productive classes. This action is permissible as none of these roads form part of a household’s essential access route to basic service centres. Better value-for-money can therefore be achieved without violating any citizen’s basic access rights. It is important to stress the need for authorities to formally unproclaim Strategic Roads, rather than simply neglecting their maintenance. Although the immediate budgetary effects of both decisions are similar for road authorities, neglecting the maintenance of a road has deadweight costs. For example, provincial education or health departments – which operate in isolation of the road departments – may decide to locate a school or clinic on a road slated for neglect. The provincial road departments, however, may be unwilling to unproclaim roads as network length is one of the factors that determine the Provincial Roads Maintenance Grant allocation. Referring to Figure 2, it is expected that most Surplus Roads will be in areas with relatively low GVA. While the structural growth model ultimately determines the classification, this indicator provides a starting point to begin rationalising the roughly 1.5 million roads in South Africa.

5 APPLICATION OF THE CLASSIFICATION SYSTEM

Once determined, these classifications can be immediately added as a decision variable in the asset management systems by pairing the road IDs with the relevant class. The intention is not to overwrite the existing classification data, as these serve unique purposes, but rather to include economic information that can help prioritise maintenance schedules. Authorities will be able to order their activities with full information about which roads they have a constitutional obligation to maintain, which structurally support high levels of economic activity, which support lower levels of cyclical economic activity, and which make a negative economic contribution. This information also compliments the engineering variables in the asset management systems, which include: visual condition; roughness; rut depth; macro texture; deflections; and traffic. Except for traffic data - which are often unavailable or unreliable, and on their own misleading - these technical variables can only prescribe remedial works. Budget allocation systems then refer to the prescribed works to allocate the limited resources in a way that minimises the whole life cost of the network. This often means roads in better condition are maintained, as delayed routine and periodic

DRAFT FEBRUARY 2018 PER – Provincial Roads – Annexure II: Working Paper 2 60 maintenance exponentially increases the total cost, and deteriorated roads are ignored. While the same strategy may still apply, the economics-based classification system ensures maintenance schedules are efficiently targeted and no important roads are overlooked.

6 CONCLUSION

The economics-based classification system effectively identifies and prioritises roads that are required to satisfy citizens’ constitutional right to access basic services and roads that maximise potential economic growth. Moreover, the system is aligned with the fundamental criteria that underpin classification systems: the classes are easy to understand and use; the definitions and identification procedures are clear; and the system is ubiquitous across the whole network. This new classification system also ensures that all roads kept in the network make a positive economic contribution. In addition, by identifying and then unproclaiming Surplus Roads authorities can limit their financial liabilities whilst not violating any human rights or ignoring their mandate. Road authorities can thus improve expenditure and investment outcomes by incorporating this information as a line item within their asset management systems.

REFERENCES: [1] COTO. 2012. TRH 26: South African Road Classification and Access Management Manual. The South African National Roads Agency Limited: Pretoria. [2] CSIR. 2017. Geospatial Analysis Platform. Online at: https://gap.csir.co.za/gap/download-maps- and-data. Accessed 10 November 2017. [3] CSIR. 2018. Urban Formal Settlements in South Africa. CSIR: Pretoria. [4] Department of Basic Education. 2017. EMIS. Department of Basic Education: Pretoria. [5] Government Gazette. 2010. Access Norms for the Department of Basic Education. Cape Town: Government Gazette 33283. [6] Intergovernmental Committee on Surveying and Mapping. 2006. Assessing Feasibility of a National Road Classification. ICSM Bi-Annual Meeting. [7] Kannemeyer, L. 2016. South African Road Network Condition, Needs and Funding. The South African National Roads Agency: Pretoria. [8] National Department of Health. 2017. Health Facility Audit. Department of Health: Pretoria. [9] National Department of Transport. 2004. Road Classification System Survey. Department of Transport: Pretoria. [10] National Department of Transport. 2006. Road Infrastructure Strategic Framework for South Africa. Department of Transport: Pretoria. [11] Ross, D., and Townshend, M. 2015. Can South African road authorities satisfy constitutionally guaranteed basic access rights without unduly sacrificing economic growth? 15th Conference on Asphalt Pavements in South Africa. [12] Ross, D., and Townshend, M. 2017. Cost-effective provision of low-volume roads in South Africa. 18th World Road Meeting in India. [13] SANRAL. 2015. Strategic Plan 2015/16 – 2019/2020. The South African National Roads Agency Limited: Pretoria.

DRAFT FEBRUARY 2018