• Abstract and Figures
• Public Full-text

Report No. 49193

AFRICA INFRASTRUCTURE COUNTRY DIAGNOSTIC

Railways in Sub-Saharan Africa

June 2009

Sustainable Development Africa Region

Document of the World Bank Vice President: Obiageli Katryn Ezekwesili Sector Director: Inger Andersen Task Team Leader: Vivien Foster

RAILWAYS IN SSA

TABLE OF CONTENTS

1 Introduction...... 1 2 The Railways...... 4 2.1 The Networks...... 4 2.2 Traffic Density...... 7 2.3 Infrastructure Condition...... 8 2.4 Network Expansion Proposals...... 11 3 Infrastructure Investment and Maintenance...... 13 3.1 Overview...... 13 3.2 Infrastructure...... 13 3.3 Economic Evaluation of Infrastructure Investment...... 15 3.4 Economics of Mechanized Track Maintenance...... 20 3.5 Indicative Investment Needs in SSA Railways...... 21 4 The Market...... 23 4.1 Overview...... 23 4.2 Traffic Trends...... 25 4.3 Passenger Traffic...... 27 4.4 Freight Traffic...... 28 4.5 Competition...... 33 4.6 Freight services...... 34 5 Institutional Arrangements...... 39 5.1 Overview...... 39 5.2 Legal and regulatory framework...... 39 5.3 Governance and management of State-owned railways...... 40 5.4 Structure of concessions...... 41 5.5 Concessionaires...... 45 6 Operational Performance...... 47 6.1 Overview...... 47 6.2 Labor productivity...... 47 6.3 Rollingstock productivity...... 48 6.4 Impact of concessioning on productivity...... 52 6.5 Service quality...... 53 7 Financial Performance...... 55 7.1 Revenue and Cost Structure...... 55 7.2 Financial results...... 58 7.3 Passenger services...... 58 7.4 Freight services...... 61 7.5 Concession financing issues...... 63 8 The Way Ahead...... 66 8.1 Introduction...... 66 8.2 Concession performance...... 67 8.3 Four key issues...... 68

i RAILWAYS IN SSA

NOTE

This report contains a number of data tables summarizing railway performance, which reflect the data which was available as at December 2008. While in several cases detailed information is available on a regular basis following privatization and concessioning, in a number of others there is far less authoritative information available compared to previously. This is due to a combination of factors: the replacement of detailed annual reports of public operators by summary reports of companies which are often diversified; the limited reporting requirements in many concessions (which even then are often only partially complied with), and the emphasis by some concessionaires on commercial confidentiality. As a result, information in many cases has to be gleaned from a variety of sources, some of which are less reliable than others. In order to provide a guide to the reliability of the information in this report, the following convention has been adopted: Where information is simply stated, it has been directly sourced from official reports or concessionaire data Where information is qualified by “it is reported that,” it has been sourced indirectly from third- party reports, the press or the internet Where information is qualified by “it is understood that,” it has been sourced from unattributable third parties. The peer reviewers and World Bank staff, particularly Pierre Pozzo di Borgo and James Leigland provided valuable contributions and comments concerning many of the transactions. However, all responsibility for the material remains that of the author.

ii RAILWAYS IN SSA

EXECUTIVE SUMMARY

The role of rail in Africa has changed greatly in the last thirty years and is likely to change just as much in the next thirty. Thirty years ago, many of the railway systems were carrying a high share of their country’s traffic, either because competing road transport had poor infrastructure or faced restrictive regulations, or because rail customers were established businesses who were locked into rail through physical connections or (if they were parastatals) through policies which directed them towards the use of a fellow parastatal. During the intervening period, both the economies in general, and transport in particular, have become liberalized. Coupled with the general improvement in road infrastructure this has led to much stronger competition and, generally, a significant loss of their previous market by railways.

Railways in sub-Saharan Africa (SSA), as in much of the rest of the world, were slow to respond to these changes. Very few governments, other than South Africa, invested significant sums of their own, or their own Government’s funds, in rehabilitating and renewing infrastructure. Such investment as there was, other than for purely mineral lines, usually came from bilateral and multilateral donors. As this typically arrived after the damage was done, and in some cases not at all, the continent is full of railways that can best be described as “walking wounded.” Whether they can be patched up by the appropriate investment medicine and a new exercise regime under a concessionaire is a moot point. In some cases, help never arrived and the railway has collapsed; this has been their fate in Guinea, Sierra Leone, the north-east network in DRC and some of the Angola short lines.

Railways have also suffered severely during the various wars and conflicts that have occurred during this period: much of the Mozambican central and northern networks, as well as railways in Angola, Ethiopia, Eritrea, Congo Brazzaville and Ivory Coast have either been damaged or have been unable to operate for long periods because of civil unrest, in some cases up to twenty years. Although it is an understandable desire of governments to reinstate networks in such cases, this is often extremely expensive and it is legitimate to question whether the transport solutions of one hundred years ago are still the most economical solution.

Much is often made of the inherent lower cost of rail as compared to road. This is certainly true where minerals have to be transported from a rail-connected mine to a rail-connected port but is not so clear-cut for medium-distance general freight which has to be transported by road to and from the railheads. Comparisons between rail and road line haul rates often show rail is much cheaper but if this were a direct measure of the value of the service then why does rail have such a small share in many of these traffics? It is because the line haul rate is only one of many factors which are taken into consideration by customers when they have a choice; the costs of pick-up and delivery also need to be considered, as do the service-level factors such as transit time, reliability, and service frequency. For rail to play a significant role in the future general freight transport system it must provide an acceptable service level and ensure it is addressing the needs of customers. In too many cases, what rail has historically offered as “transport” has been a totally different product from what the competing road hauler has been offering, and for which road is able to charge a significant premium.

The role of rail has therefore changed significantly in recent times. Within ten to twenty years, any remaining monopolies for general freight will have gone and the only traffics for which African railways, as a mode of transport, will have an undisputed grip will be mineral traffics. Historically, many such pure mineral rail transport operations have been run as an internal ancillary operation by mining companies and this is likely to be a continuing trend in the future

iii RAILWAYS IN SSA

(either through a subsidiary or a contract organization). Experience in many countries around the world has repeatedly demonstrated that general freight transport has become an increasingly dynamic business, in which operators need to be flexible, responsive and capable of adapting to changing circumstances. Fewer and fewer customers are fellow parastatals who are effectively directed to use the railway and few government-owned organizations, no matter how “corporatized” they may be, have the commercial freedom to operate effectively in a fully competitive environment, in which transport services increasingly need to be tailored to the particular requirements of individual customers. If rail is not to die a lingering death, it must adapt to the new market and become a transport business – and the predicaments of the remaining government-owned railways show it cannot compete effectively in doing this whilst it is handicapped by the bureaucratic constraints and the lack of commercial incentives and accountability of a government organization.

Since 1993, several governments in SSA have responded by concessioning their systems. Although results have been mixed, many systems have increased their traffic volumes and there has often been significant investment for the first time in many years. In general, railways have performed more efficiently and there has been little evidence of any monopolistic behaviour. However, relations with governments have often been fraught, especially concerning passenger services, and it is clear many governments (and possibly advisors) had unrealistic expectations of the difference the private sector could make in terms of both business operations and investment. In particular, it is clear that concessionaires are reluctant to spend anything on infrastructure from their own funds beyond what is required for day-to-day maintenance. Thus the funding of long-term asset renewal and upgrading remains an open question for most of the SSA network; unless this issue can be addressed, it will be impossible for rail to survive in the long-term, other than for large mineral movements, as the competition from the road networks continues to increase.

iv RAILWAYS IN SSA

1 INTRODUCTION

The changed role of rail in Africa over the last thirty years has seen it move from a situation where many of the systems were carrying a high share of their country’s traffic to one in which their market share has declined, their assets have steadily deteriorated, their quality of service has reduced, and they are in many instances only a minor contributor to solving the transport problems of the continent.

The first railways south of the Sahara were built in South Africa in the 1860’s and 1870’s, with lines heading inland from the ports at Cape Town and Durban. The networks in what were then Cape Province, Natal and Transvaal continued to develop but it was not until the turn of the twentieth century that large-scale railway development began in other parts of the continent.

In almost every case, the pattern was the same, with isolated lines heading inland from a port to reach a trading centre or a mine, and a few branch lines then being built over a period of time. As almost all the lines were constructed under colonial administrations, many of the lines were State-owned but several were also constructed as concessions or, in the case of some mineral developments, by the mining company as an integral part of its mining operation.

This process has continued until recent times, with several lines having been built since the Second World War. Although there have been grand masterplans for over a century (Figure 1), most of the African networks remain disconnected lines, either within a single country or linking a port and its immediate regional hinterland. The only true international networks are those centered on South Africa and stretching north to Zimbabwe, Zambia and DRC and, to a lesser extent, the old East African Railways network in Kenya, Uganda and Tanzania.

This also reflects the limited amount of inter-country trade. While the countries were European colonies, there was naturally little trade between, say, English and French colonies but, even today, trade volumes between adjacent countries are still often remarkably small. For example, between 1996 and 2000, less than 6 percent of Tanzania’s trade by value was with Kenya, and about 2.5 percent with her other neighbors (Zambia, Rwanda, Burundi, DRC, Uganda and Malawi); even South Africa only represented 7 percent. The same pattern can be replicated in many other countries. It can be argued that this lack of regional trade is a product of the transport infrastructure inherited from colonial times but the similarity in the products exported from many countries suggests that, even if such inter-regional links existed, it is likely they would be only lightly-used.

This pattern of economic development has meant that African railways, more than almost anywhere else in the world, are closely linked to the ports (indeed, much of Africa had integrated port and railway organizations for many years) and, where railways traverse more than one country, freight traffic is generally largely transit with comparatively little originating or terminating in the intermediate country1.

1 Other than the special case of Spoornet, the most significant exception is Kenya, one of the few coastal countries where the main port is not also the commercial capital; the railway was originally built to access Uganda but Kenya has since developed to such an extent it is now by far the larger traffic generator.

1 RAILWAYS IN SSA

Figure 1 What might have been – the Trans-Sahara and Cape-to-Cairo Railways

Some of the railways were struggling financially from the start but they generally managed to operate reasonably successfully up to the 1960’s. However, as the road system developed and larger trucks were introduced, the higher-value general freight was gradually captured and rail traffics increasingly comprised bulk mineral and agricultural traffic and semi-bulks such as fuel. Whilst this has in many cases provided enough funds to cover working expenses, railways have rarely been able (or allowed by government where they had the potential) to collect enough reserves to fund asset renewal; this has almost universally been provided on an intermittent basis through loans from multilateral or bilateral agencies, often leaving railways with a patchwork collection of disparate kinds of equipment and rollingstock. The steady degradation of the asset base has meant that even when, as in recent years, efforts have been made by railways to capture higher-value traffics such as containers, the quality of service has been so low that they have only achieved a limited market share wherever there is road competition.

Another problem for most African railways has been the continued requirement to operate passenger services without budgetary compensation. These have been required by governments but are generally covering only a part of their working expenses; this not only consumes cash that should be being used to renew the freight and infrastructure assets but also, for many railways, ties up traction power that could be being used for cash-generating freight services.

The final difficulty for many operators has been the impact of the many wars and civil disturbances that have occurred over the last fifty years. Railways are often one of the first targets for destruction and this has affected many railways, either directly (e.g. Angola,

2 RAILWAYS IN SSA

Mozambique, Ethiopia, Eritrea) or indirectly by cutting inland railways off from their ports (e.g. Malawi and Burkina Faso).

As a result, most of the railways that have been presented for concessioning in Africa have been (and generally still are) badly run-down, requiring substantial rehabilitation of both infrastructure and rollingstock. Even where they may have significant traffic volumes by local standards, these are generally low by world standards (a railway carrying more than 1 billion net tonne-km is the exception rather than the rule in much of Africa), and the concessions often come with requirements to continue operating a loss-making passenger service

Nevertheless, the rhetoric accompanying some of the transactions suggests that many politicians believe, or want to believe, that the concession award will be the prelude to very substantial investments by the concessionaires, particularly in infrastructure. To date, this has barely materialized, with most infrastructure improvements being done with international financial institution (IFI) or donor funds. The main issue for most sub-Saharan railways is whether concessioning is just a temporary solution or whether some alternative approach is needed to ensure a long-term future for railway systems providing acceptable levels of service.

3 RAILWAYS IN SSA

2 THE RAILWAYS

2.1 The Networks

At the end of 2008, there were 52 railways operating in 33 countries in sub-Saharan Africa (Annex 1 and Figure 2.1). Most of these used either the ‘Cape gauge’ (1,067 m or 3’6”) or the meter-gauge. The main interconnected network in southern and central Africa is Cape-gauge, as far north as DRC and southern Tanzania; it is also used in the ex-British possessions of Ghana, Nigeria and Sudan and, a little surprisingly2, in Congo Brazzaville. Meter-gauge is used in all the other ex-French possessions and also in the East African network linking Kenya, Uganda and northern Tanzania as well as the disconnected Ethiopian line. There has been a number of narrow-gauge lines at various times; most3 of these remaining are either derelict or not operating. There are also a number of isolated standard-gauge lines; those in Mauritania and Guinea are privately-operated mineral lines whilst that in Gabon, although primarily developed for mineral traffic, is a public railway which also carries general traffic and a passenger service. Although the multiplicity of gauges suggests that inter-operability is a major problem in Africa, this is far from the case at present as there are only three places, two in Tanzania and one in Guinea, where there are two gauges in the same location. However, if some of the proposed connecting lines are constructed, this will become more of an issue. The South African-based Cape-gauge network, in theory4, connects eleven countries, and the East African network directly connects three. There are two international meter-gauge networks in West Africa connecting land-locked French-speaking countries to the coast: Ouagadougou- Abidjan (Sitarail), linking Burkina Faso to Côte d’Ivoire and Bamako-Dakar (Transrail), linking Mali to Senegal and one in East Africa, linking Ethiopia to Djibouti. Other networks do not cross international borders but provide railheads from which traffic can be on-carried by road: Cotonou-Niamey (OCBN), which is entirely within Benin, provides a link to Niamey through a railhead at Parakou; Camrail provides railheads for traffic between the port of Douala in Cameroon and the Central African Republic and Chad.; and, in East Africa, TRC and KRC, carry traffic for Burundi (and eastern DRC) and Rwanda respectively. In some countries, parts of the network are not currently operated, either through war damage, natural disaster or general neglect and lack of funds. The total network size is around 70,000 km, of which about 55,000 km is currently being operated. Almost all the network is single-track, except for sections of the Spoornet5 network. Very little is electrified outside South Africa (where 42% of the network, or nearly 9,000 km is electrified); the only other electrified sections are 858 kilometers in DRC, (24% of the SNCC network) and 313 km in Zimbabwe, of which only the former is currently operating. The bulk of the network was built during the colonial period at the end of the 19th century and the first third of the 20th century. Since then very few lines have been constructed outside South Africa and its immediate neighbours. The most significant is the Tazara line linking Tanzania and Zambia, built by the Chinese during the 1970s; other major projects have been the Trans- Gabonais, opened in 1987, principally to transport minerals, the extension of the Cameroon network from Yaounde to Ngaoundere and the north-eastern extension of the Nigerian network from Kuru to Maiduguri.

2 This was because so it could link, presumably by ferry, with the Cape-gauge networks in DRC 3 The Eritrean network is the only one currently in operating condition. 4 Damage from civil wars means there are not currently connections to Malawi and Angola. 5 Although Spoornet recently changed its name to Transnet Freight Rail, its previous name has been retained for convenience.

4 RAILWAYS IN SSA

Figure 2.1 : Rail Map of Sub-Saharan Africa [replace with 2008 map]

Not surprisingly, the railway network density in Sub-Saharan Africa is low. The highest density (measured as route-km/000 km2) is in South Africa (16) but most other countries are in the range 1 to 6, and thirteen SSA countries currently have no operating railway at all. Too much should not be read into this statistic; network density is strongly affected by the pattern of population and China, Canada, Russia and Australia, all with large undeveloped and sparsely- populated areas, also have densities of between 5 and 7 while most European countries range from 20 to 100. More telling is the network density per million of population, which is highest in Botswana ([660]) and Gabon (520), followed by South Africa (460). Most other SSA countries

5 RAILWAYS IN SSA range from 30 to 150 km per million. In comparison, European countries range from 200 to 1000, with Australia and Canada reaching over 1500, although China is much lower at 50. However, these broad statistics in isolation are poor guides to the need for network expansion. The construction of new lines needs a minimum level of traffic for it to represent an economical transport investment and the internal geographical distribution of potential traffic generators within a country and the absolute level of usage that can be expected are more important than national averages. With over 50 companies operating on around 55,000 km of track in SSA, there are many small operators, especially as a single company, Spoornet, represents about 40 percent of the operating network and 70 percent of the traffic. Figure 2.2 shows the distribution of network length and passenger and freight traffic between the main system groups:  Southern: South Africa, Botswana, Namibia, Angola, Swaziland, Madagascar, Mozambique, Zimbabwe, Zambia and Malawi  Central Africa: DRC, Congo Brazzaville, Cameroon, Gabon  East Africa: Tanzania, Kenya, Uganda, Sudan, Ethiopia, Djibouti, Eritrea  West Africa: Nigeria, Benin, Togo, Ghana, Ivory Coast, Burkina Faso, Guinea, Senegal, Mali and Mauritania Figure 2.2 Network size and traffic by region

100% 40

90% 35 )

80% 0 0

0 30 ( 70% n o i

t 25

60% a r e p

50% o 20

n i

40%

m 15 k - e

30% t

u 10 o

20% R 5 10% 0% 0 Route-km Passenger-km Net tonne-km Southern Africa Central Africa East Africa West Africa Southern Africa Central Africa East Africa West Africa General Mineral Regional share of network and traffic Route-km in operation

14 140

12 120 ) ) n 10 n 100 o o i i l l l l i i b b ( (

8 80 m m k k - - r - e e g 6 n 60 n n e o t s

t s e a 4 40 N P

2 20

0 0 Southern Africa Central Africa East Africa West Africa Southern Africa Central Africa East Africa West Africa Non-urban Suburban General Mineral Passenger transport Freight transport Data sourced from AICD database.

6 RAILWAYS IN SSA

Box 2.1 Mineral railways in SSA SSA has a number of railways that are primarily for the carriage of coal and minerals, either for export or as an integral part of a production process, most of which have been constructed to higher technical standards (axle-load, alignment) than the conventional network:  the Mauritania line is a standard-gauge line dedicated to an iron-ore mine and was built as an integrated project  there are two standard-gauge lines and one meter-gauge line in Guinea carrying bauxite from mines to ports, also built as integrated projects. There is also a standard-gauge line linking to the meter- gauge government railway.  a Nigerian standard-gauge line has been planned and under construction for 30 years - one leg is intended to run from an iron-ore mine to the steel plant and the other from the plant to the coast. It remains disconnected from the main network but is now intended to be part of a new network linking the main cities in southern Nigeria with the capital in Abuja.  the Gabon standard-gauge line was built as a substitute for a cableway/ metre-gauge line over which exports were made through Pointe Noire in Congo Brazzaville. The dominant traffic is from the mine, which is also currently the concessionaire, but the line also carries some timber and provides a passenger service.  the Sishan - Saldanha Cape-gauge line (Orex) is dedicated to iron-ore exports but is part of Spoornet  the Richards Bay line (Coalex) is also a Spoornet Cape-gauge line, dedicated to coal exports. Some of these lines also carry general traffic for the mine. Passengers often travel informally on freight wagons and one or two have reportedly found an old passenger coach.

There are some specialist mineral lines (see Box 2.1) in both West and South Africa; these total only 4 percent of the network but carry over half the freight (as measured by net-tonne-km), most of which is carried on the Spoornet coal and ore export lines. Southern Africa dominates rail freight, handling 74 percent of the freight traffic (some of which is also minerals and coal) on the non-mineral lines and over 80 percent of the total net tonne-kilometres. Southern Africa also dominates the passenger business, with over 70 percent of passenger-km, largely because of its heavy commuter passenger business. Some other SSA cities also operate (or have operated) commuter services but, with the exception of Dakar, these are generally one or two peak-hour services a short way out along the main line (Box 2.2). Lagos, however, is currently tendering for private operators of two ‘light’ 6 rail lines to be built which are together expected to carry nearly two million passengers per day, rather more than the South African services in total. 2.2 Traffic Density

Traffic densities on SSA railways (expressed as traffic units7 per route-kilometer) are generally low (Figure 2.3)8; excluding Spoornet, the highest network average is on Gabon (2.7 million), with Cameroon (1.1 million) the only other railway to be over 1 million and many railways averaging under 300,000. By comparison, the average traffic density of the Maghreb systems (Morocco, Algeria and Tunisia) is nearly 2 million while in Europe most systems are within 2-5 million, with densities below 1 million found only in Albania and Montenegro.

6 As these are planned to be trains of eight cars or more operating on at high-frequency lines with signaling, these will be conventional commuter railways or surface metros by most measures. The infrastructure will be constructed by the government while the operator provides the trains and signaling and operates the service. 7 The traffic units carried by a railway are the sum of the passenger-kilometers and the net tonne-kilometers. It is a simple standard measure which is widely used, although it has some limitations as an indicator (e.g. a first-class passenger-kilometer in a TGV is treated identically with a passenger-kilometer in a crowded suburban train). The relative weighting of passenger and freight is conventionally taken as 1:1, although alternative weightings have been used on some railways from time to time, usually trying to reflect relative costs. 8 For clarity, this figure excludes Spoornet, which averages around 5 million overall although this reduces to only 2.4 million when the specialized coal and ore lines are excluded

7 RAILWAYS IN SSA

Box 2.2 Commuter railways in SSA By far the largest commuter rail networks in SSA are in South Africa, where Metrorail operates extensive EMU services in Pretoria, Johannesburg, Cape Town and Durban, each carrying around half a million or more commuters each day, and much smaller loco-hauled operations in Port Elisabeth and East London. In total it carries over 500 million paying passengers each year. Metrorail was operated as a distinct business unit within Transnet until 2006, when it became part of the South Africa Rail Commuter Corporation (SARCC). It has a fleet of 4200 carriages (about 70 per cent of which are operational) and runs services over more than 2000 route-km, some of which it owns and some of which is Spoornet’s. Also in South Africa, a concession has been awarded for a standard-gauge rapid (160 km/hr) regional line between Johannesburg and Pretoria. Construction began in 2006 and is due for completion in 2010/11 by an independent organisation. The only regular commuter service operating outside Southern Africa for many years has been the ‘Petit Train Bleu’ (PTB) in Dakar, which has operated since 1988 between Dakar and Rufisque, on the main line of what is currently the Transrail concession. Service frequency is relatively high, with 19 pairs of trains each weekday which are reported to carry 25,000 passengers per day. It is operated by the Agence Nationale de Nouveaux Chemins de Fer. ‘Commuter’ services in other African cities have generally been small-scale, typically one or two loco- hauled return services per day (into the city in the morning and return in the evening). Examples include Nairobi (on three routes), Lagos (two routes), Accra (two routes), Harare (two routes), Bulawayo, Luanda (one route with 6 return services daily), Maputo and Kinshasa (one route each). Annual patronage is typically in the low millions. There have sporadic attempts and proposals in some other cities (such as the Njanji service in Lusaka) but few have survived in the long-term. There are, however, now signs that changes are at last occurring: in addition to the Lagos plans mentioned in the main text, a new service was inaugurated in Kaduna in 2008 and Accra has also ordered new DMUS for its suburban service.

Even in South Africa, there are many lines with low densities; Spoornet classify the lines in their network into three groups:  high-density; those carrying over 2 million net tonnes p.a.  light-density: those carrying under 2 million net tonnes p.a. but with prospects for growth; and  low-density: those carrying under 2 million net tonnes p.a. and with no prospects for growth Around 50 percent of the network falls into the first category, 25 percent into the second and 25 percent into the third. The significance of this classification lies not so much in the distribution of these lines within the network but rather the categorization of all the lines carrying below 2 million tonnes p.a. as low-density. This would include all public railways north of Bulawayo other than that in Gabon. African railways are therefore mostly lightly-used by world standards and many networks can be expected to struggle to generate enough funds to maintain and renew their infrastructure as it is required. 2.3 Infrastructure Condition

Most African networks outside South Africa still operate with the standards to which they were constructed; some limited upgrading has occurred but the lines can still be characterized as relatively low axle-load, low-speed small-scale under-capitalized networks which are ill-suited to modern requirements. In addition, many structures, and even some of the trackwork, are now over 100 years old. Combined with chronic under-maintenance over a long period of time, many sections of track have deteriorated, almost to the point of no return. While this can be tolerated on low-volume feeder lines, and indeed may be the only way some can be viably operated, it is

8 RAILWAYS IN SSA a major handicap when competing against the modern road networks which are increasingly being constructed in major corridors. Figure 2.3 Network traffic density (average 2001 – 5)(1)

3500 STATE - OWNED CONCESSIONED )

n 3000 o i l l i m

( 2500

m k - e

t 2000 u o r

r

e 1500 p

s t i n

u 1000

c i f f a r 500 T

0 N S T B S D C C G E B D N F G K T C S Z T U B N M M a C a a r u o w t e e i e a R F o R i a g a a h a a t a h g z n m a d i t n n m n m a c M E b a l d a r C C i n s e a a o z r a a i y a n g o n z a b w n a s r i r w

r a p a b n l d o i i a n a ( ( r i a r a i a l a a n S C i l a i a a i a i a a l n i i n i N F a l l a d M C C K ) )

Passenger Freight

(1) Some railways in the ‘concessioned’ group were only concessioned close to or after 2005. Most systems have considerable sections of track that require repair or replacement, with some countries having major sections which are not in operation and will require rehabilitation before any operations can recommence: typical examples include Angola (69 percent), Benin (23 percent), and Uganda (91 percent). In other countries, many sections of the network (up to 60 percent in Ghana) do not see regular operations. Even where services are operated, poor track condition forces speed restrictions over long sections, resulting in a loss of railway competitiveness and rolling stock productivity. The cost of such repairs (conservatively estimated at $200,000 per kilometer in the most straightforward cases and probably closer to $350,000 per kilometer as an average) is quite beyond the financial capacity of most railways at current traffic volumes; funding such repairs would absorb all operating surpluses for very many years, by which time another backlog will have appeared, and so on. However, rehabilitating many of the sections in disrepair is unlikely to be economically warranted unless there are good prospects for bulk traffics or there are no alternative road connections. This is discussed in more detail in Section 3 but in summary few lines carrying less than 1 million net tonnes p.a. are likely to warrant major rehabilitation and lines carrying under 250,000 tonnes p.a. probably cannot support even anything more than routine maintenance.

9 RAILWAYS IN SSA

Box 2.3 Track structure on TRC Figure 2.4 shows the age and weight of rail on the TRC network in Tanzania in 2001, just prior to the concessioning process beginning and when the mainline was carrying about 2 million gross tonnes, a respectable figure for such a railway. Over half the network still had the original rail, nearly 90 years old and at 56lb/ft, only just capable of carrying a 15 tonne axle- load (and that only if the track was properly ballasted with sleepers in good condition – the official axleload limit at the time was 14.2 tonnes). Much of this was track was on branchlines but there was over 500 kilometers on the two main lines, heavily restricting the wagonloads that could be carried from end-to-end (which was most of the traffic).

1200

1000

800 m k - e

t 600 u o R

400

200

0 <10 10-30 30-50 50-70 70-90 Age of rail (years) Rail weight (lb/ft) 45 50 55/6 60 80 Figure 2.4 Age and weight of rail – TRC 2001 TRC had been renewing its main line over the previous ten years, financed by an IFI and donor loans, and this had allowed them to begin the process of improving their track standard by installing 80 lb rail which can easily support an 18-tonne axleload; the track renewal included new steel sleepers, adequate ballast and welded track. However, after ten years, they had still only replaced about 25 percent of the main line.

When the Dakar-Senegal railway was concessioned, the average age of track was reported as 37 years in Senegal, and 51 years in Mali. Such averages hide many problems, with much track inevitably being much older and often too light for even the moderate axle-loads9 currently being operated (Box 2.3). In addition, the strength of rail manufactured 60 or 70 years ago is often well below current standards for a similar weight of rail, leading to fatigue failures and rail fractures. Conditions in Angola and Mozambique illustrate the additional difficulties of rehabilitating railways in countries emerging from a conflict, with most infrastructure destroyed and the first task being to remove mines from the railway right-of-way. Signalling on many networks still relies in many cases on manual systems, whether with mechanical signals or through train orders. On most lines, the low train density means that these are quite adequate from a capacity viewpoint but there are often significant safety problems because of human error. Where power signaling is installed, it is often not operating due to short circuits, lack of electrical power, and dilapidated cable networks. Telephone exchanges in many companies are similarly obsolete, with limited capacity and requiring spare parts which are virtually impossible to find.

9 An approximate rule-of-thumb is the permissible axle-load is about 25% of the rail weight in lbs/yard (or 50% of the rail weight in kg/m); but in practice the limiting factor is often the bridges rather than the track.

10 RAILWAYS IN SSA

In summary, most SSA railways are confronting major infrastructure problems associated with:  Aging track: insufficient ballast; rail wear (especially on curves), deteriorating earthworks and formation  Civil work: most structures are in poor condition.  Rail signaling and telecommunications: obsolete equipment and lack of spare parts The total amount required to overcome these problems is very large. In the case of TRC (Box 2.3), the cost to upgrade the remainder of the main line was about USD 300 million at that time. Its gross revenue was about USD 60 million p.a. and it broke even on working expenses (i.e. excluding depreciation), only by deferring maintenance of infrastructure and rollingstock. Perennially deferred maintenance leaves a harsh legacy which is generally beyond the capacity of a railway to self-finance and the only option in most cases is seeking large concessional loans and/or grants from third parties. 2.4 Network Expansion Proposals

There have been many proposals, some dating back a century, to create new routes for landlocked countries and integrate the isolated networks. The most ambitious was in 1976, when the African Railways Union (ARU) prepared a masterplan for a pan-African rail network which included 18 projects requiring 26,000km of new construction (Annex 5), many of which had been proposed for several decades. This plan, designed to create a grid to support intra- African trade development and regional economic integration, was approved by the OAU in 1979 but few, if any, of the proposed SSA links have got beyond the drawingboard. The ARU is now concentrating on a 2001 revised masterplan containing a subset of ten corridors (Table 2.1)10, in some of which the network is already partially constructed (e.g. Corridor 8). Table 2.1 ARU ten-corridor masterplan Corridor Countries linked 1 North – Centre – South Libya-Niger-Chad-CAR-Congo-DRC-Angola-Namibia 2 West – Centre Senegal-Mali-Burkina Faso-Niger-Nigeria-Chad Senegal-Mali-Burkina Faso-Niger-Nigeria-Ghana Côte d’Ivoire-Ghana-Togo-Benin-Nigeria- Cameroon 3 North – East Sudan-Ethiopia- Kenya-Tanzania-Uganda 4 North – East – West Sudan-Chad-Nigeria 5 East – South Tanzania-Rwanda-DRC-Uganda Dar Es Salaam-Kigoma-Burundi 6 East – Centre Sudan- Central African Republic-Cameroon Kenya-Uganda-DRC 7 North Morocco-Algeria-Tunisia-Libya-Egypt-Mauritania 8 East – South Tanzania-Zambia-Zimbabwe-Mozambique-South Africa 9 Centre – South Cameroon-Gabon-Congo- DRC-Angola-Namibia 10 North – West Senegal-Mauritania-Morocco There have been a number of regional studies and action plans for subsets of these corridors in SSA:  in West Africa a major feasibility study for network development was launched by the African Development Bank (ADB) in 2004. Other studies have looked at connecting Benin-Niger-

10 In 2005, the ARU further simplified this with its adoption of three major trans-continental routes:  Libya – Niger – Chad – CAR – Congo – DRC – Angola – Namibia (6500 km).  Senegal – Mali – Chad – Djibouti (7800 km).  Kenya – Tanzania – Uganda – Rwanda – Burundi – DRC with possible extensions to Ethiopia and Sudan (5600 km)

11 RAILWAYS IN SSA

Burkina Faso-Togo (the Africa Rail project of 1070 kilometers costed at $US 6 billion) as well as the rehabilitation of existing lines such as OCBN)  preliminary studies have been undertaken by the UN Economic Commission for Africa (ECA) in both Central Africa and East/South Africa to improve linkages  in 2000, COMESA launched the Great Lakes railway project to improve connections between the Great Lakes and the Southern African rail network. This project involves both rail and water transport on Lakes Tanganyika, Kivu and Edward connecting Burundi, DRC, Rwanda, Uganda and Zambia. It includes a new rail link from the Zambian network to Mpulungu on Lake Tanganyika, a series of five lines linking to the Ugandan network, rehabilitation of the Kasese line in Uganda as well as modernization of six lake ports.  a project promoted by the Northern Corridor Transit Transport Coordination Authority, based in Mombasa, to link Kisangani with Mombasa with a new line from Kasese to Kisangani, with feeder lines linking Kasese to Goma and from there to Kigali and Bujumbura via Bukavu. More recently, there have been a number of proposals for individual lines, most of which are either segments of the corridors in Table 2.1 or (in the case of the TransKalahari) of the original 18 links given in Annex 5:  a link from Isaka in Tanzania to Rwanda together with complementary links from Rwanda and Burundi to the Uganda and Tanzanian network  a link through Kenya (or possibly Uganda) to southern Sudan  extending the Lilongwe line in Malawi from Mchinji to a railhead at Chipata in Zambia (with the aim of then connecting with Tazara to provide an east-west trans-Africa link once the Benguela line has been rehabilitated  Namibia has been promoting (and partially constructing) routes from Walvis Bay to Zambia and Angola to develop the port hinterland There have also been a number of proposals for dedicated mineral lines by mining companies. Prior to the end of the minerals boom, the coal deposits at Moatize in western Mozambique were being developed by the Brazilian mining company, CVRD11. Although the line from there to Beira is being upgraded as part of the rail concession awarded in 2005, its planned capacity is relatively low and CVRD have examined an alternative of constructing a new high-capacity line to Nacala, which is also a much better port. This line would be a dedicated mineral line, probably standard-gauge, with no interaction with the networks it crosses. Similar dedicated minerals lines have also been planned in other parts of Africa e.g. the Pepel line in Sierra Leone (a rehabilitation of a line abandoned over 20 years ago) and the Chinese line in Gabon. The TransKalahari railway from Botswana to Walvis Bay has been regularly suggested over the past twentyfive years, primarily as a route to export the Mmamabula coal deposits in Botswana via Gobabis and an upgraded Namibian network through the port at Walvis Bay. The total distance to the port is 1480 km, of which 880 km would be new construction. Whilst there is no shortage of schemes, the economics of many of the proposals are uncertain and even the proposed mineral railways are likely to be postponed for several years following the recent slump in commodity prices. The next section examines the likely investment required and the extent to which it can be economically justified.

11 This company operates major mineral railways in Brazil in its own right.

12 RAILWAYS IN SSA

3 INFRASTRUCTURE INVESTMENT AND MAINTENANCE

3.1 Overview

The dividing line between ‘investment’ and ‘maintenance’ is far from well-defined in railways. The term ‘investment’ has historically been used both for new construction and rollingstock, for replacement rollingstock and also (but not universally) for the rehabilitation and replacement of track when it has become life-expired. In some cases, where railways have been chronically short of funds and have relied on third parties for funding, the periodic overhaul of locomotives and rollingstock has also been classified as ‘investment’. In addition, there are often choices that can be made whether to maintain assets in their condition when constructed (as far as that can be done) or to let them steadily deteriorate and then decide whether they should be replaced; this is particularly so given the long life of much railway infrastructure, for which demand may now be very different to what was envisaged when it was originally constructed. Railway assets have a finite life, which can be measured in terms of either years or usage. For the low-tonnage lines common over much of Africa elapsed time is generally the determining factor although usage can be a factor in certain areas, such as wear on rails where the track is curved. However, whether driven by time or usage, asset life will be shortened if maintenance is not carried out on a regular basis, as has unfortunately often been the case on SSA railways. This is exacerbated as assets become obsolescent; one of the causes of under-maintenance on many African railways is the difficulty of getting spare parts for equipment which has long ceased manufacture. This is a particular problem for communications and for non-mechanical signaling systems12 but diesel locomotives also often present difficulties; whereas steam locomotives had relatively few components and parts could generally be mended or manufactured locally, most diesel locomotives need a regular supply of imported spare parts. These problems are exacerbated when there is a plethora of different makes and models acquired under a range of donor programs13. Providing a detailed estimate of the investment currently needed by SSA railways is a daunting task. Besides detailed inventories and assessments of infrastructure condition, a view also needs to be taken as to how much investment is economically justified; lines which have been, or will be, superseded by road developments, or those with low traffic levels, will rarely merit reconstruction and investment funds should instead be directed to those parts of the network with a long-term future. Although it is an understandable desire of governments to reinstate such links, this is often extremely expensive and it is legitimate to question whether the transport solutions of one hundred years ago are still the most economical approach. The volume of investment required at any one time to maintain a system in good order is therefore a function of the age and condition of the existing assets; where it has been under- maintained, as is the case with most SSA railways, a substantial volume of ‘backlog’ investment will be required to bring it back to a condition ‘fit-for-purpose’. It also depends on the extent to which life-expired assets should be replaced and, if so, whether this should be on a like-for-like basis or whether, for example, track standards such as axle-load should be progressively upgraded14. 3.2 Infrastructure

Infrastructure is generally the largest investment item for a railway. The construction cost of a new single-track non-electrified railway in relatively flat terrain is at least $US1.5 million per kilometer, increasing to $5 million or so in more rugged country requiring more extensive

12 In one notorious case, spare parts were no longer available as little as ten years after the system had been installed. 13 At the time of its concessioning, Tanzania had 96 operating locomotives of 10 different types from eight different manufacturers. 14 Many African railways were built with axle-loads of around 15 tonnes; replacement programs should generally consider increasing this to 17 or 18 tonnes, at least on mainlines..

13 RAILWAYS IN SSA earthworks. Reconstructing an existing line, for which the right-of-way and earthworks already exist, will typically cost about $350,000 per kilometer if new materials are used and rather less, say $200,000 per kilometer, if secondhand material such as cascaded rail can be obtained. Where lines are to be upgraded, bridges may need to be strengthened to handle higher axle- loads; additional earthworks may also be required if alignments are to be improved (as this is almost always in the sections through hillier country). Overall, a reasonable rule-of-thumb is that upgrading can easily double the cost of simple track renewal. These costs exclude signaling, for which much relatively cheap options are now available for the typical low-volume network. During its life, track will normally be subject to both annual maintenance as well as periodic maintenance of various types. A well-run railway will inspect its network at regular intervals; large railways use a track recording vehicle, and these can also be used by small railways if they are connected to a larger adjacent system (e.g. the Southern Africa networks could use one from South Africa). This will normally be done once or twice a year for lines carrying around 2 million tonnes or less. Heavily-used lines also normally inspect the rail itself using specialist equipment but this would be not normally be required on most SSA networks. Besides routine maintenance, including attention to earthworks and drainage, which is normally done continuously, track that is used on a regular basis is normally tamped to ensure the track geometry remains within the limits required to operate trains safely at the nominated speed. This is done at intervals ranging from annually (or more often if the track is very heavily used) to every six or eight years; the actual frequency depends on the tonnage using the line as well as engineering factors such as the type of sleepers, whether the track is welded or not and the condition of the subgrade on which the ballast has been laid. For low-volume lines tamping might be done every four or five years but where the track structure is weak (e.g. there are a large proportion of rotted sleepers) and volumes are very low (say under 250,000 net tonnes p.a.) the decision is often taken not to tamp at all but to let the track gradually deteriorate and to impose speed restrictions as necessary. This strategy is adopted on many low-volume branchlines in Australia as well as on many of the ‘short’-lines in the US (see Box 3.1). Eventually, the various elements of the track structure wear out. This is generally caused by a combination of usage and elapsed time; timber sleepers generally have the shortest life but this is not a major issue on many SSA lines where steel sleepers have been used almost since initial construction. Track which has been relaid with concrete sleepers will have an equally long life – fifty years is a commonly-quoted figure as long as the concrete specification was adhered to. Rail life is normally directly related to usage; even though there is an order of magnitude difference between the life of rail on sharp curves and that on straight track, on most SSA

Box 3.1 ‘Short-lines’ in the US and track quality Although the seven Class 1 rail lines in the US normally dominate discussion of US rail operating practices, there are also a large number of secondary railways (or ‘short-lines’) which handle local traffic and also act as feeders to the Class 1 systems. In 2007 these secondary railways included 33 ‘regional railroads’ (these own over about 500 km of track and have revenues greater than US$40 million) and 523 ‘local’ railways. These latter are further classified into ‘shunting and terminal’ railways and local line-haul railways which do not meet the regional railway criteria. The regional railroads typically operate about 800 route-km and carry an average traffic of 3-4 million tonnes; the local linehaul railroads operate about 150 route-km and carry an average of about 400,000 tonnes, similar to the smaller SSA networks. Overall, the average traffic density is about 2 million tonnes. Track quality standards in the US are defined by the Federal Railroad Administration; this classifies track into categories ranging from Class 1 (sidings and yards where the maximum speed is set at 15 km/hr) to Class 9 (mainline on which speeds of up to 350 km/hr are permitted). Most track on the short- lines is in Classes 1 – 4: the regional railroads have about 55 percent of their track in or below Class 2 (maximum speed 40 km/hr) while the local line-haul railways have over 70 percent below Class 2 and only 4 percent in Class 4 (where the maximum speed is over 60 km/hr and under 80 km/hr).

14 RAILWAYS IN SSA railways rail life will be measured in decades rather than years. Finally, ballast should also have a reasonably long life as long as it was of good quality to start with, although over time it will gradually degrade. These lives all assume that the track was properly built to start with, with good-quality materials. Unfortunately, many of these railways were built with an eye to economy; formations were not properly consolidated, ballast was skimped and so on. Whilst these problems were manageable in the early days when rail had an overwhelming advantage over other modes of transport15, they became more serious as its monopoly was challenged and particularly since it has needed to upgrade itself technologically to survive. Under-strength formations inexorably lead to high maintenance costs and reduced operating speeds and eventually there are no options other than to reconstruct the line or to cease operations. When such infrastructure has been reconstructed it should have a life of at least 40 - 50 years, given the generally low traffic volumes; the cost of periodic reconstruction is thus equivalent to an annual cost of $5000 - $10000 per kilometer. (This excludes any return on investment; if a 5 percent ROI is included, the annual cost increases to $20,000 - $40,000 per kilometer16). A line carrying 1 million net tonnes p.a. will thus need to earn US 0.5 - 0.8 c/ net tonne-kilometer (ntk) to fund the reconstruction but traffic on one with a density of 250,000 tonnes p.a. will need to find US 2-3 c/ntk. If ROI is included these numbers quadruple at a minimum. As yields on most freight railways in Africa are around US 4-5 c/ntk, this means that only those lines with a density of 2-3 million net tonnes or more can even begin to consider full rehabilitation from a purely commercial viewpoint unless the investment creates a significant increase in traffic. Although this is a relatively high threshold, from an economic perspective, there are significant benefits, in addition to those accruing to operators, which also need to be considered. These are discussed in the next section. 3.3 Economic Evaluation of Infrastructure Investment

The criteria, economic or otherwise, for track and infrastructure investment, should be related to the policy framework within which the railway operates17. This typically has three main elements. Firstly, the railway can play an important economic role in transporting freight more efficiently than can other modes, using fewer resources and creating smaller external impacts on the environment and general public. This is can be true over relatively short distances for bulk freights such as ores and minerals, petroleum and agricultural products, over medium distances for semibulk traffics such as steel and cement and over longer distances for general freight. When greater efficiency flows through to the manufacturers and consumers of the traffics being transported, this will then often create more general developmental benefits. Secondly, the railway can often perform an important social role in carrying passenger traffic that could only be transferred to road at a higher cost in road infrastructure, accidents and environmental impact. However, as these services rarely, if ever, cover even their variable operating costs in Africa, any expenditure and investment associated with these services needs to be concentrated on those services where rail is most competitive. Finally, as rail is competing for both recurrent and capital expenditure with other transport modes, and with other sectors, investments should be targeted to ensure it is operated as efficiently and effectively as possible. Against this policy background, the key economic criterion for the selection of track and infrastructure investments is the same as for any other investment – that the economic benefits

15 When the Gold Coast (later Ghana) Railway was built, its rates were about half those by cart and between one-sixth and one-half of those for head-loading. 16 For a commercial ROI of investment of 10-15 percent, these numbers would double or treble again. 17 This section draws heavily on the discussion found at www.worldbank.org/wbsite/external/topics/extratransport/extra????

15 RAILWAYS IN SSA of the investment are greater than the economic costs and that projects should be prioritised according to the relative size of these benefits, normally as measured by the economic internal rate of return (EIRR). When track is rehabilitated or upgraded, benefits accrue not only to operators through reduced operating costs but also to two other groups:  users of the infrastructure who benefit from improved services  non-users of the infrastructure who benefit from the reduced impact either directly (e.g. reduced congestion on roads) or indirectly (e.g. from economic development made possible by a new line or more general effects such as a reduction in greenhouse gas emissions) User benefits Service improvements arise not just from reduced linehaul tariffs but also from improvements in service quality such as transit time, reliability, damage and security. Whilst rail generally (but not always) has a lower linehaul cost than road, it also often has additional pick-up and delivery costs which need to be considered; road also generally has a faster transit time, is more flexible in terms of scheduling and is often more reliable. The economic benefits of infrastructure investment are therefore a function of the overall service level rather than just the linehaul tariff; this can readily be seen in the many situations where rail linehaul tariffs are well below those of competing modes but where rail has only a small share of the market18. It also explains why rails strength in many countries is in low-value bulk commodities, for which there are often direct rail connections, where marketing is relatively simple and where level-of-service issues are relatively straightforward. It is also why freight volume, throughout Africa, has generally shown an increase following privatisation; tariffs have not reduced but the level of service has improved considerably. Establishing the user benefits associated with any particular scheme therefore involves not just a comparison of linehaul costs but also these other attributes of travel. They are collectively known as generalised cost and user benefits are calculated as the difference in generalised cost between the ‘with-investment’ and ‘without-investment’ cases19. There is an extensive literature discussing the relative importance of the various components (cost, time, reliability etc) for passenger traffic but rather less for freight traffic. Where an existing line is already in operation, the impact on demand of relatively small changes in the level of service also can be measured using elasticities, which measure the response in demand to a change in the level of service. There is an extensive literature on these as well, although again more for passengers than for freight. Non-user benefits Non-user benefits arise as a result of the change in transport flows by both road and rail. Where the investment enables rail to attract traffic that previously moved by road, there are three main non-user benefits, all related to reduced use of the road network:  Reduction in the level of road congestion  Reduction in road damage and consequent savings in road maintenance costs  Reduction in road accidents

18 In an extreme case, at one time some of the Fiji sugar-cane tramways were free but still could not compete with road on the overall level of service. 19 Where the overall volume of traffic is expected to change because of the investment, the benefits attributable to the incremental traffic are normally valued using what is known as the ‘rule-of-a-half’. A description of this can be readily found in the literature.

16 RAILWAYS IN SSA

In addition, there may be more general benefits such as a reduction in the level of emissions, including those of greenhouse gases (GHG), depending on the particular circumstances of the project. Congestion savings will be project-specific; clearly if the parallel road network is uncongested, (say, with a V/C ratio under 0.6-0.7), any such savings will be small but they are often very significant in urban areas, particularly in the approaches to ports. Road damage costs are broadly a direct function of tonnes carried, with an extra weighting for overloaded vehicles20. The precise cost depends on parameters such as the standard of initial construction, the level of overloading and the mix of traffic but a typical figure for a sealed arterial road would be 0.5-1 c/ntkm. (with lower figures for the better-constructed roads which are also generally the more densely-trafficked ones). Road accident costs are disproportionately affected by the number and size of heavy vehicles on the road. Precise accident rates by type of vehicle are difficult to establish but are likely to be at least 1 fatality per 1 million heavy-vehicle-km. Converting this to an economic cost requires assumptions about the a ‘value of life’ (VOL). Estimates of this differ widely from country to country but a survey in 2000 concluded that VOL was typically valued at about 120 times GDP/head, giving a value of about US$50,000 for a country with a per capita GDP of US 400. Combining this with the assumed fatality rate gives an accident cost of about 0.5 c/ntk. Rail freight is about 15-30% as damaging as articulated trucks with respect to greenhouse gases (GHG) in terms of emissions per net tonne-km (g/ntkm). Typical values of CO2-equivalent are 4 g/ntkm for rail and 15-30 g/ntkm for articulated and rigid trucks respectively. GHG are subject to a wide range of valuations, but US$10-25 per tonne is a commonly-used range; this implies a benefit from rail of around .02 – 0.1c/ntkm. Other road vehicle emissions (especially CO, NOx, CH4 and particulates) are reducing substantially as new standards are introduced; their most serious impact is increased mortality, particularly from NOx, hydrocarbons and particulates, but reliable estimates of how changes in emissions convert to changes in mortality are still under development. Operator benefits Operator benefits from track and infrastructure improvements derive from the better quality of track, allowing higher operating speeds and reducing infrastructure maintenance costs, rollingstock maintenance costs and fuel consumption. Figure 3.1 demonstrates the impact on overall track quality of different track structures and tamping strategies. A modern tamper applied to a satisfactory track structure produces a cyclic pattern of steady deterioration in track quality (along AB), until tamping occurs, at which point the original quality is recovered (BC) and the cycle recommences. However, if the existing track structure is weak e.g. through lacking ballast of either the right quantity or quality, the track quality deteriorates faster than should be the case, along the path A...B3...C3, requiring a shorter tamping cycle. If a railway is not able to tamp at B3 due to shortage of resources, the track continues to deteriorate until it reaches the point B4, at which point it is retamped. Rollingstock and routine track maintenance costs steadily increase during this time because of the steadily declining track quality and its reduced capacity to absorb loading, and eventually speed restrictions of increasing severity are imposed.

20 More precisely, a measure known as standard axles, which heavily weights overloaded vehicles. Vehicles that are 30% overloaded

17 RAILWAYS IN SSA

Figure 3.1 Alternative tamping strategies

B6 NOT TO SCALE Speed restrictions

B2

B4 (Worse)

B5 Minimum acceptable track quality

B3 B

B1

Track quality A1 C1

C3 C A

Modern tamper, improved track structure Modern tamper, existing track structure Target track quality (Better) Existing tamper, existing track structure Existing tamper, existing track structure, poor sleepers

Elapsed time

The lack of track quality is compounded if obsolete tampers are operating (or manual tamping is being performed), both of which normally produce lower standard track than is possible with a modern tamper, following the path A1...B1...B2 before retamping occurs. If stretches of track also have sleepers in poor condition, the deterioration in track quality in these cases is even more rapid, along the path A1...B5...B6. This cyclic pattern of a steady deterioration in track quality, which is then recovered by tamping, either partially or fully, continues until the track becomes life-expired, it is no longer possible to significantly improve its quality for any length of time and a progressive decline sets in. Once deterioration begins in life-expired track, it rapidly accelerates and severe speed restrictions have to be imposed. At the same time, routine track maintenance costs increase and rollingstock costs rise because of the reduction in track quality. Although the general process underlying track deterioration is well understood in general terms, track deterioration functions vary significantly between different railways and lines (primarily because of variations in subgrade strength and the standard of initial construction). However, what is certain is that failure to undertake periodic maintenance on a well-used line will shorten the life of its track to a very significant extent. The benefits of track rehabilitation vary from system to system and depend on both the existing and new track structure that is being installed, as well as the volume of traffic handled over the line. Major maintenance benefits can be expected from replacing jointed track with welded track

18 RAILWAYS IN SSA and the improved track quality following rehabilitation will reduce wear on rolling stock bogies, wheels and drawgear, typically by 5-10% and reduce fuel consumption by 2.5 - 5%. The major operating benefit is the increased operating speed with its consequent savings in all time-related costs, particularly those associated with rollingstock utilisation. Indicative evaluations Although the economics of any specific track investment will depend on the level of traffic and its likely pace of deterioration if the investment is not undertaken, some order-of-magnitude evaluations of track renewal have been summarised in Table 3.1 to provide guidelines as to which projects are likely to be realistic candidates for investment. The evaluations are based on the following assumptions:  A track rehabilitation cost of $250,000 per km (i.e. some cascaded or re-used material)  A range of ‘with-project’ traffic volumes from 0.25 to 5 million tonnes  Only 50% of the forecast ‘with-project’ traffic will be carried in the ‘without-project’ case  Average road tariffs of 9c/ntkm and average rail tariffs (which are assumed to cover full costs) of 4c/ntkm  Average pick-up and delivery charges of $100/tonne for an average haul of 500 km (equivalent to a 2c/ntkm penalty against rail)  A service quality penalty of 10% against rail (i.e. if the road door-to-door rate is within 10% of the rail door-to-door rate, users will choose road because of its better service qualities)  Rollingstock maintenance and fuel savings for existing traffic of 7.5% and 4% respectively  External benefits as derived above (0.75 c/ntkm for road maintenance, 0.5 c/ntkm for road accidents and 0.1 c/ntkm for GHG). Table 3.1 Example EIRR for track rehabilitation Volume p.a. Benefits ($ 000 pa) EIRR (mill net tonnes) User Operator External Total (% pa) 0.25 1 3 2 6 0 0.50 3 5 3 11 2 0.75 4 7 5 16 5 1.00 6 9 7 21 8 1.50 9 12 10 31 12 2.00 12 16 14 41 16 5.00 29 39 34 101 37 The examples show that, under the assumptions adopted, a volume of around 1-2 million tonnes needs to be expected before major track rehabilitation is economically worthwhile. Traffic volumes under 1 million net tonnes will provide poor returns and the strategy for such lines should be to look for cheaper approaches than full rehabilitation – using cascaded rail and with partial sleeper replacement, for example, is all that can be justified. For very low-volume lines, where no more than 0.5 million tonnes could be expected, the case for anything other than ‘patch-and-mend’ is likely to be weak. These conclusions should be treated as indicative rather than being absolute guidelines. If the line is carrying bulk traffic, then user benefits can be expected to be rather greater, as the pick- up and delivery penalty of 2c/ntkm will probably no longer apply – and user benefits will increase significantly (in this example by 100%). Similarly, if the local road network is very poor, road maintenance savings are likely to be rather greater than those assumed and hence external benefits from rail rehabilitation are likely to be greater. Table 3.1 shows the benefits in these examples accrue equally to users, operators and externally, as it is assumed 50 percent of the traffic will travel by road if the investment is not undertaken. If some of it will instead be suppressed, and not travel at all, the external benefits

19 RAILWAYS IN SSA will reduce accordingly. More generally, ‘with-project’ traffic forecasts need to be carefully reviewed. The service’ penalty’ of 10% assumed in this example is conservative – it assumes an efficiently-run service-oriented railway which can probably only be achieved in practice through concessioning. A penalty of 30%, which is probably closer to the mark for the typical government railway, will reduce the user benefits by 60%, as well as eliminate most of the general traffic that might be carried. Anything that can be done to improve service levels will thus improve both forecast traffic as well as economic returns disproportionately. This example excludes any congestion benefits, which are highly location-specific, but these would also need to be included in any evaluation which involved, for example, rail access to and from ports through heavily congested urban networks. While individual projects should all be evaluated according to their particular circumstances, the general conclusions of Table 3.1 are likely to remain true. In particular, low-volume lines (say below 0.5 - 1 million net tonnes p.a.) are very unlikely to merit full rehabilitation and lines carrying bulk traffic will normally generate greater (and more certain) benefits than those carrying general traffic. 3.4 Economics of Mechanized Track Maintenance

The complete renewal of the track itself may only arise every thirty to forty years on most African railways but during that life most mainline track, in addition to the routine maintenance that is carried out by track gangs, will also require periodic maintenance at regular intervals. This consists of realigning the track geometry within the tolerances required for the desired operating speed i.e. ensuring the track is in a tolerably straight line and also that it does not have undulations, whether going uphill, downhill or on the flat. Historically this was done by gangs of men but most railways, especially those which have installed concrete sleepers, now use specialized equipment called tampers. The main objective of tamping is to eliminate, as far as possible, irregularities in the ‘line’ and ‘level’ of track and thus allow trains to operate at the design speed for each section of track. This in turn:  reduces transit times through the increased line speeds, increases track capacity and improves rolling stock productivity  reduces track and rolling stock maintenance costs and fuel consumption through the improved track condition Under normal circumstances, tamping would generally take place on SSA mainlines every 3-4 years or so (equivalent to AC in Figure 3.1). However, if machinery is not available or the track structure is substandard, it may need to be undertaken every 2-3 years. If this is not possible due to lack of resources, track quality is above the ‘minimum acceptable’ for 12-24 months every cycle and temporary speed restrictions often need to be imposed, especially if the sleepers are also in poor condition. Reasonable estimates of the impact of not tamping are that, similar to track rehabilitation, the wear on rolling stock bogies, wheels and drawgear increases by 10% and fuel consumption by 5%. It will also increase recurrent track maintenance costs, possibly by up to 20%, compared to what will be achieved by using the new equipment and increase transit times and thereby operating costs. Tamping typically costs $3000 per kilometer, including materials, but will generate financial benefits of around 0.1-0.2 c/ntk as well as some savings in unscheduled maintenance. Whether it is worth doing or not therefore depends on the volume of traffic using the line in question. Typical financial evaluations show that the return to the operator for lines carrying 500,000 – 1 million net tonnes is good, with rates of return between 50% and 100% but that for volumes

20 RAILWAYS IN SSA below 500,000 tonnes the returns are much more meagre and below 250,000 net tones they are no-existent. The external benefits that can be attributed to tamping are not likely to be as great as those related to track renewal and a rational SSA railway manager would therefore be unlikely to tamp anything other than his mainlines, with the objective of extending their lives as long as possible. In practice, any such decision will be complicated by two factors: passenger services normally require a higher track quality than freight in order to provide an acceptable ride quality and mechanized tamping is only possible in practice if a machine is available (not always the case on isolated networks) and if the track structure is strong enough to withstand the process. The first will lead to more frequent tamping than would otherwise be the case for a pure freight service whilst the second will require geometry correction to be done manually, almost inevitably to a lower standard than is possible using mechanized equipment. 3.5 Indicative Investment Needs in SSA Railways

A full analysis of the investment needs of SSA railways would require detailed data on infrastructure and rollingstock condition and traffic volumes for each railway. In the absence of this data, an indicative estimate has been made using aggregate statistics and broad assumptions supported by the previous discussion. The SSA network north of South Africa consists of about 44,000 km of track; about 34,000 km of this is currently operational. Nearly all the lines are low-volume by most definitions and would thus only justify partial rehabilitation, possibly using cascaded materials. Even assuming a relatively low unit cost, say $200,000 per kilometer, probably no more than 15-20,000 kilometres of the network can justify this level of expenditure. Spreading this over an average 40-year interval, the cost of infrastructure rehabilitation would thus average around $100 million p.a. Rollingstock should cost about $80 million p.a. The cost of replacement rollingstock can be estimated in a similar manner, using assumed average asset lives. The SSA network north of South Africa carries around 15 billion ntk p.a. (excluding the mineral lines) and about 4 billion passenger-km (pkm). This will require, on average, replacing 500 wagons, 20 passenger carriages and about 20 locomotives each year. As with infrastructure, many of these in practice will be second-hand, from India or from South Africa but the estimated cost will still average about $80 million p.a., equivalent to about US 0.4 c/ntk or pkm. The steady-state investment in the SSA network north of South Africa (allowing $20 million for facilities, maintenance equipment and so on) should thus be around $200 million p.a. The backlog investment is possibly up to $3 billion. The amount required at the present moment is much larger as it is a function of the backlog that has been allowed to accumulate; assuming a 15-year backlog would give an additional one-off requirement of about $3 billion, although in practice this expenditure could be spread over, say, a 10 year period at an annual rate of $300 million. The combined annual program would then be about $500 million over ten years, after which investment would reduce to the steady-state level of $200 million. These calculations only provide broad order-of-magnitude estimates. However, it is clear that the total amount required to overcome these problems is very large, of the same order as the annual revenues of some of the railways and well beyond their capacity to self-finance; the only option in most cases is seeking large concessional loans and/or grants from third parties. In addition to re-investment in the current network, there is also the possibility of investing in completely new projects. There have been many proposals, summarized in the previous section. Few of these proposed projects will be financially or economically viable. Whilst there is no shortage of schemes, In many cases, the new routes are competing with existing road and rail routes and hence the rates that could be charged would be constrained, typically to at most US 5c/ntkm. In the case of export mineral traffic, the potential rate is also generally constrained to around US 2-3 c/ntkm by the long-term delivered market price. As a serviceable two-lane

21 RAILWAYS IN SSA road can generally be constructed for around $1 million per kilometer, the additional rail investment will be economically justified only if the forecast traffic volumes are typically at least 2-4 million tonnes p.a.; if, however, the capital costs of the infrastructure do not have to be recovered, the lines can probably be operated successfully at rather lower volumes, say 0.5 – 1 million tonnes21. These ‘top-down’ estimates can be cross-checked against ‘bottom-up’ estimates available for some railways. Investment plans have recently been developed in Ivory Coast, covering the next 20 years, and Madagascar, covering the next decade. Infrastructure investment is equivalent to $8-10,000/route-km p.a. and rollingstock is US 1-1.4 c/ntk; around 50 percent of these amounts are planned to be provided by governments. In the US, where most investment is privately-funded, the short lines typically invest about $6,000/ route-km p.a. on infrastructure; the Class 1 railways, with the far greater average density of about 30 million tonnes, invested $45,000/route-km in 2007.

21 The experience of the Alice Springs – Darwin railway supports this conclusion. Although it was constructed at an average cost of under US1 million per kilometer, with 40% funded from government grants, it has never carried more than around 2 million tonnes p.a. and in 2008 went into administration. It was able to cover its working expenses but could not service the 40% percent of the capital cost which was debt-funded, with the equity shareholders writing off their investment.

22 RAILWAYS IN SSA

4 THE MARKET

4.1 Overview

Most railways in SSA carry far more freight than passengers (Figure 4.122). During the eleven years from 1995 to 2005 freight traffic represented about 80 percent of total traffic units. However, almost all railways carried some passenger traffic; only Swaziland and, since 1998, Uganda are freight-only railways. There is also little difference in aggregate between the concessioned railways and the non-concessioned railways. However, as freight volumes have either grown, or at least been maintained, the proportion of traffic that is passenger is steadily reducing and by 2005 was down to 18 percent. In only a few cases, and these are mostly railways which have very low traffic levels overall and have almost completely lost their freight business, does the percentage of passenger traffic reach 50 percent. Several of the other railways with a reasonable passenger business (e.g. TRC, Tazara and Transrail (prior to 2005)) only do so because competing road networks either do not exist or are in very poor condition. Figure 4.1 Passenger and freight traffic on SSA railways (annual average 1995-2005)(1)

2000

1800 STATE - OWNED CONCESSIONED

1600

) 1400 n o i l l i 1200 m (

s t i 1000 n u

c i 800 f f a r

T 600

400

200

0 N S T B S D C C G E B D N F G K T C S Z T U B N M M C a a a r u o w t e e i e a R F o R i a g a h a a a t h g a z n a m n d t n i m n m a c a E b M l d a C r C s i e n a a z o r i a y a a g n n o a z n b w r a s i r w

r a p a b n l o i d a n a i ( ( r i a r a a i l a a n S C i a l i i a a a i a a l n i i n i N F a l l a d M C C K ) )

Passenger-km Net tonne-km

(1) Some railways in the ‘concessioned’ group were only concessioned close to or after 2005; this figure should therefore not be used to compare the impact of concessioning on traffic volumes Figure 4.1 also highlights the general lack of scale of most SSA railways. The busier railways typically carry one billion traffic units; Spoornet carries this volume of traffic every three days. Of course, the Spoornet network is much larger but in overall terms most SSA railways have similar volumes to a moderately busy branch line. In some cases this is due to lack of demand but in many cases it is caused by shortages of rollingstock, particularly locomotives.

22 This figure covers all railways in Africa other than South Africa (because of its relative size), Zimbabwe (because of lack of data on passengers in recent years), and the Senegal and Mali internal traffic (the international traffic was concessioned to Transrail and is included in the chart). Some data is missing for some countries so the averages are indicative rather than precise and all data is subject to the caveats discussed in Box 4.1. The data used for the chart is given in Annex 1.

23 RAILWAYS IN SSA

Box 4.1 Traffic and operating statistics Almost every railway in the world maintains traffic and operating statistics and there are several sources generally available e.g. the UIC website provides a long time-series for many countries and the World Bank also has a database. Most of the government railways also publish (or used to publish) annual reports which included a summary of the key indicators. However, these sources often disagree and there are many traps for the unwary. Reported traffic statistics may differ because of what is included; freight may or may not include traffic carried for the railway itself such as ballast and fuel (this can sometimes be very large, one year in Nigeria it was over 90 per cent of the traffic) or traffic hauled by third parties (such as SEFICS in Senegal) while passengers may or may not include staff and other persons using passes (such as politicians and war veterans), commuter passengers carried in trains operated under contract to other authorities and so on. Other problems include the date of the statistics, a particular problem when the reporting year ends in June, say, and data can thus be in different years in different sources. Less easily spotted problems arise from the method of compilation; tonnage statistics can be derived from revenue accounts or from operating statistics – the two rarely agree to within 5-10 percent even on the best-run railways; passenger statistics often include season tickets converted to trips based on ‘multipliers’, factors which convert a weekly ticket, for example, to ten trips (say) – and sharp jumps are sometimes seen when assumptions are changed. And finally some problems arise from simple issues of arithmetic, data entry (in the case of websites) and presentation; when concessions started towards the end of the year, in at least one case (and possibly two) the remaining months of the calendar year were included in the following year’s results; thus the reported traffic for the last ‘year’ before the concession was actually eleven months of results while that for the first ‘year’ of the new concession was actually thirteen months. Thus, while the order of magnitude of these statistics is generally correct, the precise value will often vary between sources. The conclusions drawn in this paper are therefore based on general trends rather than detailed results for individual railways except for selected cases where this data has been extracted especially for the purposes of analysis.

The average haul on SSA networks (Figure 4.2) is relatively long in terms of their network size but not especially long in terms of competing with road. Some railways carry predominantly end- to-end traffic; TRC, Tazara and Transrail all have an average distance for freight of around 1000 kilometers and some smaller railways, such as Uganda or the Mozambique lines, act as feeders to other systems which subsequently on-carry the traffic a further few hundred kilometers. These systems have a good chance of competing for general freight traffic, even as the road network is improved, as long as satisfactory service levels can be achieved. However, the shorter systems which require transshipment to road at railheads will generally find they can only compete effectively for bulk traffics where these exist. Few of the systems are operating significant commuter services and the average distance of passenger trips in Figure 4.2 thus generally represents trips made between (generally) the capital city of the country concerned and major provincial centers. The only significant cross- border flows are on the Sitarail, Transrail and Tazara networks.

24 RAILWAYS IN SSA

Figure 4.2 Average distance traveled on SSA railways (annual average 1995-2005)(1)

1600

1400 STATE - OWNED CONCESSIONED

) 1200 m k (

e 1000 c n a t s

i 800 d

e g

a 600 r e v

A 400

200

0 N S T B S D C C G E B D N F G K T C S Z T U B N M M a a a C r u o w t e e i e a R F o R i a g a h a a a t a h g z n m a n d t n i m n m a c M a E b l d a C C r n s i e a a z r o i a y a a g n n o z a b n w r a s i r w

r a p a b n l o i d i a n a ( ( r i a r a a i l a a n S C l a i i a a i a i a a l n i i n i N F a l l a d M C C K ) )

Passenger Freight (1) Some railways in the ‘concessioned’ group were only concessioned close to or after 2005. 4.2 Traffic Trends

The growth or decline of traffic on many systems over the last decade has often had little to do with changes in the underlying demand. In some cases, war or natural disaster has had a major impact: thus Sitarail and Congo Brazzaville both experienced sharp reductions in traffic during civil wars and CEAR suffered badly when a major bridge was destroyed and it took over two years for funds to be mobilized23 for its repair; TRC also suffered severely from another cyclone in 1997. Conversely, when infrastructure has been rehabilitated, as is happening in both Madarail and on the Sena line (part of the CCFB concession) traffic will increase sharply from a low base. In several other cases, the volume carried reflects the availability of rollingstock, particularly locomotives24. Many of the fleets are old, with spare parts not always available. Many SSA railways have a low locomotive availability and, when this is improved either through new or secondhand locomotives being obtained, or through a locomotive rehabilitation project, traffic increases accordingly. Figure 4.3 indicates the general trend in traffic over the last decade, comparing average traffic levels for passenger and freight for 2001-5 with the averages for 1995-200025.

23 Under the concession agreement, the repair of a structure destroyed by an Act of God was the responsibility of the Government, which had no funds. 24 This situation also occurs at peak periods on much larger railways e.g. Spoornet came under heavy criticism in 2007 and 2008 for not having sufficient capacity to carry export coal and minerals to ports but the situation on many SSA railways is more fundamental and long-term. 25 Abnormal years (wars, cyclones etc) have been excluded from the averages.

25 RAILWAYS IN SSA

Figure 4.3 Traffic growth on SSA railways (annual average 2001-2005 compared with 1995 - 2000)(1)

180% 0

0 160% STATE - OWNED CONCESSIONED 0 2 - 5

9 140% 9 1

e 120% g a r e

v 100% A : 5

0 80% 0 2 - 1

0 60% 0 2

e

g 40% a r e

v 20% A 0% N S T B S D C C G E B D N F G K T C S Z T U B N M M a r a C a u o w t e e i e a R F o R i a g a h a a a t a h g z n m a n d t n i m n m a c M a E b l d a C C r i n s e a a o z r y a i a a g n n o a z b w n a s r i r w

r a p a b n l o i d a n a i ( ( r i a r a i a l a a n S C a i l i a i a a i a a l n i i i n N F a l l a d M C C K ) )

Passenger Freight (1) No comparable data available for FCE, Transrail and Madarail. Some railways in the ‘concessioned’ group were only concessioned close to or after 2005; this figure should therefore not be used to compare the impact of concessioning on traffic growth During this period, most SSA countries experienced steady economic growth. GDP grew on average at 4 percent p.a., with corresponding increases in trade, and per capita GDP by about 1.5 percent p.a.; countries, such as Tanzania, Mozambique and Mali, which avoided political upheavals grew up to 50 percent faster. In spite of the generally favorable economic background, only five railways have seen an increase in both passenger and freight traffic over the period, three of which (Gabon, KRC and CEAR – which grew in spite of the cyclone damage) have been concessioned. The two non- concessioned railways in this group are Namibia and the special case of SNCC, recovering from civil war. Only one other railway has seen an increase in average passenger traffic; this is Botswana, but this also saw a drop in its freight traffic as through traffic between South Africa and the north was rerouted via the new Beit Bridge Railway. All other railways saw a reduction in passenger traffic, and a good many in freight traffic as well, including Sudan, Nigeria, Benin and two of the three Mozambique railways. The Zambia concession (RSZ) also falls into this group, although this was caused in part by a deliberate withdrawal from local intermine traffic for a period. Figures 4.4 and 4.5 show detailed trends in passenger and freight traffic in four railway companies (Sitarail, Camrail, CEAR and RSZ), before and after concessioning. Freight traffic has generally, but not universally, increased. Civil war badly affected Sitarail, and CEAR was affected by the bridge collapse. Passenger traffic has generally stagnated or declined, especially in the case of RSZ. Sitarail has subcontracted the passenger service and now only provides a haulage service. Figure 4.4 Passenger traffic Figure 4.5 Freight traffic

26 RAILWAYS IN SSA

400 1200 Date of 350 concessioning Date of 1000 concessioning

300

) Civil war in Ivory ) n o m Coast from late- i 800 l k l t i 250 Civil war in Ivory 2003 n

m ( n

Coast from late- o i m l

2003 l k i

- 200 600 r m e (

g t n Sitarail h e 150 g i s

subcontracts e s 400 r a passenger F P 100 service Rivi Rivi bridge Rivi Rivi bridge collapse 200 collapse 50

0 0 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 Years since concessioning Years since concessioning

Sitarail Camrail CEAR RSZ Sitarail Camrail CEAR RSZ

4.3 Passenger Traffic

Passenger rail services worldwide often serve two distinct markets:  transport of suburban passengers where a suburban network exists: given the high population densities in large metropolitan areas in Africa, rail offers a way to counter urban traffic congestion and improve mobility between city centers and suburban areas.  intercity transport: regional and long-distance transport linking major centers to rural areas The suburban services currently operating in SSA are summarized in Box 2.2. The only service outside South Africa that provides a reasonable all-day frequency is the Petit Train Bleu, in Dakar and even this only operates on one route. All other suburban services are essentially morning in-bound and evening out-bound loco-hauled services with few, if any, services at other times. Several cities have plans for the introduction of modern commuter networks; based on experience elsewhere in the world, these services will all need substantial external funding, both for capital and recurrent operating costs, and should be operated by new and independent transport authorities separate from the existing railway, as in South Africa. Railways were historically the only practical mode of transport in many countries for longer- distance land travel. Governments controlled fares to what were considered affordable prices; these were almost invariably below cost – in the best cases they just covered the costs of train operation but normally contributed little to either infrastructure costs or the capital cost of replacement rollingstock. Overall railway finances could only be balanced by contributions from the freight business. As the road networks have improved, strong competition is now provided by buses and shared taxis in many corridors26 in terms both of price and of service frequency and few corridors remain in which rail passenger services are the only effective means of transport. Formal compensation schemes, such as Public Service Obligations, have been introduced in a few cases to support the rail services but these rarely provide timely compensation for operating the services. Payments from the government may take several years or otherwise take the form of a subsidy calculated to break-even in overall terms, limiting the ability of railways to increase their maintenance and negating any attempts to improve the financial performance of the freight services. As a result, most long-distance passenger services in Africa are trapped in a cycle of minimal investment, deteriorating services, declining patronage and financial losses. There are generally two, and sometimes three (when sleeper accommodation is offered), classes available on passenger services but passengers are overwhelmingly (80 – 90 percent)

26 When the Tanga – Moshi road was rebuilt in the mid-1990’s virtually all the then train passengers transferred to buses and the passenger service was withdrawn within months.

27 RAILWAYS IN SSA third-class (Figure 4.6); few as they are, most first-class passengers in practice are either railway or government employees. Most trains therefore have one or two upper-class carriages, often with low occupancy levels, with as many third-class carriages as are either available or as required to accommodate demand, often well into double figures. Load factors on many trains are therefore often quite good: in Tanzania the average third-class load factor was around 70 percent27 during this period. Figure 4.6 Passengers by class

First Second Third First Second Third

100% 100%

80% 80%

60% 60%

40% 40%

20% 20%

0% 0% Passengers Revenue Passengers Revenue

Tanzania 2001 Botswana 2002 Average passenger tariffs typically range from 1-3 US cents/passenger-kilometer (Figure 11) but on most routes, medium and upper-income passengers have moved to car and air while buses, where they operate, generally, but not always, provide strong competition for economy passengers (see Section 4.5). Passenger services also carry parcels and small freight; these typically add about 25 percent to the revenue from passengers but even so revenue is generally insufficient to cover the cost of running the train (see Chapter 7). 4.4 Freight Traffic

Freight traffic on SSA railways is dominated by bulk and semi-bulk commodities (Figure 4.7), principally to and from ports. Of the railways included in the figure, only Botswana has a large proportion of non-port traffic, with most flows either supplying raw materials or receiving product (cement, petroleum etc) from South African manufacturers. The actual commodities transported by rail reflect the economic structure of countries served by the railway, with mining products (copper, tin, manganese, stone and coal) important in several countries and timber important in West Africa, together with export crops (cocoa, coffee, cotton, cereals). Imported flows are mostly manufactured products such as cement, petroleum products and general freight. On some systems much of the general freight is containerized and higher value cash crops (such as coffee from Uganda) are increasingly traveling in this way, particularly when the trip involves crossing an intermediate border before the port.

27 This is the dynamic load factor i.e. passenger-km: seat-km. As not all passengers travel to the end of the route, this means occupancy is generally much higher at the maximum load point.

28 RAILWAYS IN SSA

Figure 4.7 Commodities carried on selected SSA railways (various years around 2001)

100%

80%

60%

40%

20%

0%

Timber Cement Fertiliser Petroleum Ores and minerals Agriculture General

Unlike for passenger services, it is common to have significant imbalances between traffic in the two directions; even where tonnage is approximately balanced, the differences in the commodities, many of which require specialized wagons, mean trains are rarely fully loaded in both directions. In some cases (e.g. export traffic from Zambia to the ports), this imbalance in traffic is accentuated for rail as road vehicles delivering imports tend to backload freight at marginal cost, leaving rail to transport the remaining freight without a compensating return load. A significant proportion of traffic on many railways is to or from a third country, either directly or by road from a railhead (Figure 4.8). Import traffic is generally rather larger than export traffic; the only exceptions are a few railways such as Gabon which have substantial export mineral flows. Railways which cross international borders tend to be dominated by international traffic. The main exception is KRC, for which traffic to and from Nairobi and other centers is much greater than to and from Uganda.

29 RAILWAYS IN SSA

Figure 4.8 Traffic mix on selected SSA railways (various years around 2001)

100%

80% e g a n

n 60% o t

f o

e g a t n

e 40% c r e P

20%

0% Botswana Zambia Ethiopia Uganda

Local Import Export Transit Figure 4.9 Share of domestic and international freight traffic on SSA railways (by tonnage(1) for various years around 2001)

100%

80%

60%

40%

20%

0% l l l i i i a a a Z E B C C C N C C R i M r a a a n F S r r r b D A R R R F D R R a i a s a z C R T E C K N m U C C t G w m n i a a C C s a a S T t r C N o T B

International Domestic (1) Because of the large differences in average haul for domestic and international traffic, the share by tonne- kilometers is rather different in many cases (e.g. Transrail, for which international traffic is 97%)

30 RAILWAYS IN SSA

Figure 4.10 Average yields for passenger and freight traffic (average 1995 - 2005 or similar)

20

18 STATE - OWNED CONCESSIONED

16 ) m k

t 14 n / c

, 12 m k

p 10 / c

S 8 U (

d l 6 e i Y 4

2

0 N S T B S D C C G E B D N F G K T C S Z T U B N M M C a a a r u o w t e e i e a R F o R i a g a a h a a t a h g z n m a d t n n i m n m a c E b M a l d a C C r n s i e a a o z r a y i a a g n o n a z b w n s r a i r w

r a p a b n l o d i a n a i ( ( r i a r a a i l a a n S C a i l i a i a a i a a l n i i i n N F a l l a d M C C K ) )

Passenger Freight (1) No comparable data available for Nigeria, FCE and Swaziland. Some railways in the ‘concessioned’ group were only concessioned close to or after 2005; this figure should therefore not be used to compare the impact of concessioning on yields Average freight tariffs over the period have typically ranged from 3-5 US cents/net tonne- kilometer (Figure 4.10)28. Except for occasional semi-monopolies such as SETRAG, tariffs are generally constrained by competition, either from road or from alternative routes. The highest rates in the figure are the two DRC railways and CFCO (Congo Brazzaville), for all of which road competition is both expensive (or impossible) and lacking security, and Uganda, which is a misleading average as it is distorted by the railferry operations, the very short distances much of its traffic moves on the port access lines and a deliberate policy during this period of equalizing tariffs on all three routes (direct rail, ferry via Kenya and ferry via Tanzania) to promote competition. Typically a range of tariffs are charged for the different commodities. All railways originally had tariffs based on the classical structure related to the value of the commodity, with low-value commodities such as fertilizers being charged low rates and manufactured goods being charged high rates29. Over the years this structure underwent some changes as particular traffics received discounts either because of a more general policy, such as supporting particular industries or agriculture or because of astute lobbying by individual interests. As road transport developed, it picked off the highly-rated commodities; rail was often slow to respond by increasing the rates for low-rated commodities, partly from managerial and bureaucratic inertia

28 Current tariffs in $US terms in 2008 were rather larger than these averages in many cases; the 2008 freight yield for Camrail was 9.7 c/ntk, for Sitarail was 6.4 c/ntk and for Transrail was 8.0 c/ntk. These may not be sustainable as fuel prices fall but it seems they will remain some way above the ten-year average. 29 This structure also reflects to some extent the relative densities of these traffics: a wagonload of coal will generally weigh much more and cost more, but less per net tonne, to haul compared to a wagonload of textiles. Low-density freight is thus normally charged a comparatively high rate per tonne to allow for this

31 RAILWAYS IN SSA and partly from a fear of losing such traffic as remained30. There have been gradual moves to ensuring tariffs at least cover marginal operating costs but many SSA railways had only a moderate understanding of their cost structure (in addition many Government railways believed all their costs were fixed except for fuel and some other directly variable out-of-pocket expenses, which they probably were given many budgeting practices). Inter-railway comparison of tariffs needs to be done with some care, as tariffs are also a function of distance (e.g. Uganda in Figure 4.10) but Figure 4.11 compares tariffs for a range of typical commodities for selected SSA railways. Figure 4.11 Average yields by commodity for selected SSA railways

10.0

9.0

8.0

7.0

6.0 m k t n

/ 5.0 c S U 4.0

3.0

2.0

1.0

0.0 Petroleum Containers Cement Fertiliser Cereals Cotton products

Sitarail Camrail Transrail Madarail TRC 2001 ZR 2003 CEAR 2003

Petroleum products and container traffic generally have tariffs significantly higher than average, while agricultural products and semi-bulk commodities such as cement and fertilizer are priced below average. These variations are due to a number of reasons:  a hangover from traditional value-based tariff structures (e.g. containers often have relatively high-value consumer goods)  a reflection of the relative cost of carrying different commodities (e.g. bulk commodities which have higher net loads per wagon are cheaper to carry while petroleum is normally carried in tank wagons which have a comparatively high ratio of gross to net tonnes and, moreover, are almost always returned empty)

 whether traffic is moving in the more lightly-loaded31 direction (often exports in SSA countries) tariffs are often much lower both for road and for rail as there is excess transport capacity  volume will also influence tariffs, with many railways negotiating contract rates with large users, generally with some conditions attached such as minimum levels of traffic and, on the

30 Another factor in some cases is that shippers by rail have faced unofficial charges for wagon supply and expediting transit in addition to the tariffs in the rate book. 31 Generally termed the ‘backloaded’ direction.

32 RAILWAYS IN SSA

railway side, some guarantee of level of service. Casual users with small volumes, by contrast, will often pay the published (or ‘book’) rates which are generally significantly higher. 4.5 Competition

4.5.1 Passenger services For many years the rail networks provided the only practical way of traveling along many centers are now linked by roads which generally enable road to provide a competing service. In some cases, rail, even with the low speeds typical of most routes, still provides a faster service than the competing bus (e.g. Yaounde – Ngaoundere in Cameroon and Cuamba – Nampula in northern Mozambique) and there are often difficulties by road in the rainy season if the road is not yet paved. However, as long as the road is maintained in reasonable condition, buses have generally taken a large part of rail’s previous traffic and are likely to take more as the highway networks are progressively improved. Passenger travel by road ranges from luxury bus on a few routes through ordinary bus to minibus or bush taxi, with a range of fares reflecting the quality of service provided. Bus fares are typically about 30 -50 percent higher than the economy rail fare but on most routes are faster and sometimes twice as fast (Figure 4.12), with generally a much higher service frequency. Many bus services suffer from the same problems as rail: often unreliable departures with buses often waiting until it has a full load of passengers, delays and breakdowns en route and overcrowded accommodation, particularly on the cheaper services which are most directly comparable to the cheaper rail services. A major attraction of rail for some passengers is the better perceived safety and, on some routes, the ability to carry large quantities of produce and baggage on trains more cheaply than by bus. Nevertheless, bus generally has the lion’s share of the market (Box 4.2). The long-term prospects for non-urban rail services are generally poor, particularly as the road network compares to improve. The cost of maintaining rail track and signaling to enable operating speeds which are even marginally competitive with an averagely-maintained sealed road (say 70 km/hr commercial speed) is significantly more than for the 30-40 km/hr commercial speed needed for a freight railway and very large capital expenditure is required to construct new medium-speed (say 200km/hr) interurban railways. Such expenditure needs substantial demand (several million passengers p.a.) as well as relatively high-income passengers to even cover the cost of its operation and there are few, if any, corridors in Africa in which such expenditure could be justified for at least the medium-term.

Box 4.2 Rail passenger market share in Botswana and Burkina Faso In Botswana in 2002, there were two trains (one day, one overnight) between Gabarone and Francistown, which each took about eight-nine hours for the 435 km trip. The connecting road was a two-lane sealed road in generally fair-to-good condition, with moderate traffic levels. There were 35-40 buses in each direction each day, which took four-five hours and left at regular intervals during the day but without any overnight services. The day rail service charged Pula 19 (economy) and Pula 37 (business) and the night service charged Pula 28 (economy), Pula 107 (business) and Pula 125 (sleeper). The competing bus fare ranged from Pula 30 to Pula 38, depending on the type of vehicle and speed. (The exchange rate at the time was Pula 5.5 = US$1). Both rail and bus provided reasonable links to the local urban bus services at each end of their jounrey. In spite of the much cheaper fares, bus had about 70 percent of the market at the time and, with the day train since taken off, its market share is now probably well over 80 percent. Between Ouagoudougou and Bobo Diassollou in Burkina Faso, there are frequent buses taking about five hours. There is a single train three times weekly, which provides a service between these two cities en route between Abidjan and Ouagoudougou. The trip by rail takes nine hours and is often heavily delayed. Perhaps not surprisingly, bus is reported to have 95 per cent of the market between the two cities.

33 RAILWAYS IN SSA

Figure 4.12 Comparison of bus and rail fares and travel times

0.8 Dar - Mwanza 0.7 Douala - Yaounde Nairobi - Mombasa

0.6 e Ouagoudougou - m i

t Bobo

l Dar - Kapiri Mposhi e

v 0.5

a Gabarone - r t

l Dar - Dodoma Francistown i a r

0.4 : e m i t

Windhoek - l

e 0.3

v Swakopmund a r t

s

u 0.2 B

0.1

0.0 0.0 0.5 1.0 1.5 2.0 2.5 Bus fare: rail fare

The local trains serving villages with no road connection pose a different problem. In Malawi, the local services are primarily used by traders bringing goods to and from regional centers and the passenger train is really a mixed train with some two or three large open wagons into which the accompanying goods are loaded. This is a highly inefficient way to bring goods to market and, even though the trains are well loaded, the revenues barely cover 30 percent of the avoidable cost. While such services can be funded through explicit payments by government, the long- term solution is to create alternative means of transport by developing feeder roads that can provide basic motorized access. This will not only enable far more cost-effective means of goods transport but also greatly improve the general level of accessibility for such locations. 4.6 Freight services

Comparison of road and rail freight tariffs, and of freight market share data, needs to be done with some caution. The economics of rail transport are significantly affected by whether the ultimate origin and destination are rail-connected i.e. whether a mine or a cement works has a rail siding and whether there is ready access to a port or power station. Transhipment between the railway and the ultimate origin and destination can be surprisingly expensive, often up to the equivalent of 200 - 300 kilometers of line-haul transport and negating any advantage rail may have in pure line-haul tariffs. New sidings are sometimes constructed but these need a minimum traffic volume to be economic for a railway. Traffic which comes from disparate origins and which needs to be collected at a central depot before dispatching by rail is thus more vulnerable to road competition; conversely minerals and other bulk traffics tend to use rail as long as the service is available with sufficient capacity. However, even bulk traffics are not immune to road competition (Box 4.3); road has also been used for the relatively short-distance intermine traffic in the Zambian copperbelt, even though the mines and processing plants are connected by a purpose-designed network. Rail has faced additional difficulties in many countries over the last twenty years or so with the general liberalization of the economy and the abolition or restructuring of the many statutory

34 RAILWAYS IN SSA agricultural organizations. These were often the only means of marketing crops and had a network of depots located at key points on rail networks to which producers brought their products for storage and subsequent dispatch. Marketing channels for such products have now become more diversified and the railways have often been slow to respond to the new structures, steadily losing market share. The deserted rows of rail-connected warehouses for export cash crops at ports such as Dar-es-Salaam are silent testimony to these changes. Inland distribution networks for consumer and intermediate products have similarly developed. Although there are still inland depots for petroleum products, the trend is for deliveries direct from main depots and refineries to end-users, with consignment sizes that are far better suited to road. General freight, whether containerized or not, is dispatched in relatively small consignments and mixed loads, with freight from two or three suppliers to the same destination, are a common occurrence. Again, the costs of pick-up and delivery for such traffic work against rail in its traditional mode and in many cases this traffic almost disappeared while railways were state-operated (Box 4.4).

Box 4.3 Minerals transported by road in Ghana and Botswana Bauxite and manganese to the port of Takoradi have been the dominant freight traffics on Ghana Railways for over a decade, representing about 90 percent of the tonnage loaded. However, most years, GRC has been unable to carry all the traffic offering due to lack of rollingstock (aggravated by poor infrastructure limiting operating speeds and causing extended cycle times) or, as in 2008, due to a strike, and it has gone by road at an additional cost of $US1 per tonne for the manganese ore and rather more for the bauxite. Figure 4.13 shows it could have carried an extra 30% of traffic if the capacity had been available.

2500

2000

1500 s e n n o t

0 0 0 1000

500

0 2001 2002 2003 2004 2005

Manganese by rail Bauxite by rail Manganese by road Bauxite by road Figure 4.13 Minerals transported by road and rail - Ghana In Botswana, a copper mine north of Francistown was producing copper concentrate which was then sent to a smelter at Selebe Phikwe (a round trip of 400km), for the last 150 kilometres of which it paralleled an existing railway. The concentrates were carried in flexi-tippers (2 x 5-axle trailers), each carrying 50 tonnes net and making up to 2 round trips/day on the public road system. Rail would have been lucky to make one round trip per day and would also have required a storage and transshipment facility. Annual tonnage was forecast to reach 450,000 tonnes p.a., unfortunately insufficient to justify constructing a connecting line to the mine.

35 RAILWAYS IN SSA

Box 4.4 General freight to and from Niger The primary export and import corridor for Niger’s products links Niamey to the port of Cotonou in Benin. However, the combined road/rail service offered by OCBN has lost much of its customers to private road carriers, reducing from 88% of Niger’s total imports in 1992 to 77% in 1998 and 34% in 2005. Niger’s key imports include petroleum products, cereals, sugar, sulphur and other containerized products. The transport of cereals highlights this trend: OCBN’s share of the traffic dropped from 98% in 1992 to 3% in 2005; for other products, such as petroleum products, or sugar, the road market share is now 100%.

Comparing freight rates and market shares between countries needs to take into account not only physical factors such as infrastructure and vehicles but also the type of freight, the direction of travel (forward-loaded or back-loaded), the prevalence of overloading and other institutional factors. Figure 4.14 compares road and rail rates for a number of corridors in or about 2003. Even though the data all refers to a 12-meter container being transported inland from a port, there is still a wide variation between different geographical areas, reflecting the varying standard of infrastructure, types of road vehicles and institutional factors such as unofficial en- route charges, border crossing procedures and the impact of the freight associations common in Central and West Africa. Figure 4.14 Indicative freight rates – 12 meter container ex port (2003)

5

4 m k

- 3 r e n i a t n o c /

$ 2 S U

1

0 West Africa Central Africa East Africa (NC) East Africa (CC) East Africa (SC) Southern Africa

Road Rail

Source: UNCTAD Development of Multimodal Transport and Logistics Services 2003 and various Some of the reasons for the regional variations are straightforward: the poor condition of roads in Central Africa and (although they are now being improved) the Central Corridor in east Africa between Dar-es-Salaam, Burundi and Rwanda, the very large trucks32 operating throughout Southern Africa and the impact of the freight associations. Rail rates in the forward direction are generally about 50% of the road rate. Part of the difference is because of the extra cost of pick-up and delivery but service quality (transit time, reliability and security) is also a major factor as in all but a very few railways in the developing

32 Much of the long-distance freight in southern Africa is carried on large double trailer, seven-axle combination rigs, with a nominal maximum GVM of 56 tonnes. Typical payloads for dense traffics such as cement or steel are 30-40 tonnes.

36 RAILWAYS IN SSA world. Rates in the backloaded direction are generally around half to two-thirds of those in the forward direction, with rail rates similarly discounted. Most state-owned railways face considerable difficulties in competing against road operators because of the constraints of public ownership. Concessioning can normally be expected to bring much greater commercial flexibility as well as sharp improvements in the level-of-service. Nevertheless, the road rates (discounted by maybe 15% for the road access legs for a typical corridor and a further 10-15% for level-of-service) will nearly always act as an upper limit in most corridors. The wide range of characteristics of freight traffics means that there is often a wide variation in the modal share of particular commodities; it is not uncommon to get market shares of 100 percent for one traffic and zero for another33. Overall, rail typically carries 20 – 50 percent of the non-mineral freight in a corridor; examples range from Mombasa - Uganda, where rail carries 19 percent to Douala – Cameroon/Chad, where rail carries over 60 percent. Some of the smaller state-owned railways have a negligible share; CDE carries around 3 percent of the traffic between Djibouti and Addis Ababa. In addition to intra-corridor competition, many African railways also face competition from other corridors. In West Africa, the inland countries (Mali, Burkina Faso and Niger) all have a choice of ports:  Bamako (Mali) can access Dakar (Senegal) (1288 km) by rail or by a relatively poor road; it can also access Abidjan (Côte d’Ivoire) either directly by road or by road to Bobo-Dioulasso in Burkina Faso and thence by Sitarail to the port. In past times, it has also used Conakry in Guinea.  Niamey (Niger) and Ouagadougou (Burkina Faso) can access Abidjan (Côte d’Ivoire), Takoradi and Tema (Ghana, by road), Lome (Togo) and Cotonou (Benin) Other examples include the Great Lakes region, which has the choice of Mombasa or Dar-es- Salaam, Malawi, which uses Dar-es-Salaam, Nacala, Beira and South Africa and Zambia, which can use Dar-es-Salaam (via Tazara) or South Africa. Some of these alternative corridors are rail-served, others are not. The road transport business in SSA has been extensively analyzed in a recent study which highlighted the differences in organization, structure, costs and prices (Figure 14) between West Africa, East Africa and southern Africa, which are reflected in the differing levels of freight rates and modal competition in the various countries. There is little doubt road competition is strongest in southern Africa, which has the most liberal market structure, the largest trucks and the best roads. Two major factors also influencing road competitiveness are the level of user charges and the prevalence of overloading. Few Governments directly charge trucks with road user fees which fully compensate for the cost of providing and maintaining the arterial road network. Requiring rail to fund 100% of its maintenance and improvements while tolerating road cost under-recovery and overloading on arterial routes may help Government budgets in the short-run but is an almost impossible handicap for most general freight railways to overcome in the medium and long-term. In summary, while most railways are generally able to carry bulk minerals when they are offered, rail must offer a reasonable level-of-service for general freight if it is to be able to compete with road without offering a significant price discount. In general, freight markets in Africa require reliable services rather than high-speed services; a two-day transit for a trip of 1200 kilometers, equivalent to a commercial speed (i.e. including en-route stops) of 25 km/hr does not require a maximum speed of much more than 60 kilometers per hour. It does, however, require ensuring the infrastructure and rollingstock is maintained ‘fit for purpose’,

33 For example, in northern Botswana rail carries 100 percent of the soda ash traffic from Sua Pan to South Africa but none whatsoever of the general freight from Johannesburg in the opposite direction.

37 RAILWAYS IN SSA operating discipline to ensure en-route stops do not disrupt schedules and commercial arrangements that ensure agreed rollingstock turnround (often a limiting factor for small railways) is achieved. Achieving such an acceptable level-of-service, combined with flexible pricing policies and a strategy of providing a transport service as opposed to merely a line-haul operation, is one of the major potential benefits that a concessionaire can offer compared to the typical state-owned railway. With such improvements, the price discount between rail and road can be reduced, increasing the contribution that freight can make to the maintenance and renewal of infrastructure. Concessionaires have shown they will take a range of initiatives to achieve this. Some are physical, such as the proposed construction by Sitarail of an intermodal terminal in Ouagadougou to service the surrounding region, and others procedural, such as the introduction by ZRS of company customs bonds to sharply reduce waiting times for import traffic at Victoria Falls. These actions all combine to make rail more competitive in the market place and ensure that its potential advantages are realized in practice. However, an additional important factor is that, increasingly, being a rail operator is not enough; instead the railway needs to provide a transport service, covering the full journey from origin to destination. Conventional railways are poorly equipped to do this, both physically and bureaucratically. They do not have the equipment and find it difficult to be sufficiently operationally and commercially flexible to respond to issues along the whole length of the logistic chain. Concessionaires have the commercial freedom to deal with such issues and railways are being increasingly operated worldwide as not an end in themselves but as a part of an integrated transport business34.

34 Note Bolloré’s classification of their railway business as an “activity connected with transport” or the mining companies, who often relegate their railways to a “surface equipment engineer.”

38 RAILWAYS IN SSA

5 INSTITUTIONAL ARRANGEMENTS

5.1 Overview

In general government organizes the transport sector in SSA, typically through a Ministry of Transport mandated to develop and implement policy ensuring transport contributes to overall economic and social development. Until the 1980s almost all African railway companies were publicly-owned corporations, with varying degrees of financial and management autonomy. The first steps to increase the level of commercialization were the introduction of Contrat plans, initially in West Africa; these defined more explicitly the relationships between governments and railway companies but governments rarely, if ever, met their obligations to public corporations’ management. A new wave of reform in the African rail sector began in the 1990s with the railway concession. In this model, the state remains the owner of some or all of the existing assets, typically infrastructure, and transfers the other assets (normally the rollingstock) responsibility for operating and maintaining the railway to a concessionaire under the terms and conditions stipulated in a concession agreement. Most countries in Central, East and West Africa have moved all or part of the way to concessioning, often under the pressure of multilateral and bilateral organizations who have until recently been the only source of large loans for asset rehabilitation and renewal. With the exception of Southern Africa (South Africa, Namibia, Botswana and Swaziland) and countries suffering or recovering from civil disruption (Angola, DRC and Zimbabwe), most other countries are at various stages of a reform process. Of the 30 SSA countries with publicly-owned railways, fourteen have opted for a concession arrangement and one operates under a management contract. A further four have begun the process (Table 5.1). 5.2 Legal and regulatory framework

The introduction of concessions has required substantial changes in the legal and regulatory framework in many countries. In the francophone countries, which inherited French legal systems, concessions could generally be done without new legislation but the English-speaking countries generally have, or should have, amended their various Railway Acts. Tanzania passed a new Railway Law in 2003 and Zambia has drafted a new one which has not been implemented. Malawi planned to amend its Act but this has never occurred. The regulatory frameworks have also had to be amended to allow for the economic and safety regulation of concessions. Historically, as railways were state-owned, economic regulation was imposed through the relevant ministry whilst the railways were generally effectively35 self- regulating for safety matters (and also regulated any independent operators on their networks). The advent of concessions meant that many governments had to consider more transparent regulatory procedures. At the same time, as concessions are almost always for a defined length of time, infrastructure (and sometimes rollingstock) was generally leased to concessionaires and successor bodies to the existing railway organizations needed to be established who would be the asset owners. In French Africa, these are the ‘patrimony’ organizations (e.g. the Société Ivoirienne de Patrimoine Ferroviaire (SIPF) in Côte d’Ivoire and the Société de Gestion du Patrimoine Ferroviaire du Burkina (SOPAFER B) and Burkina Faso) with parallel bodies being set up in some, but not all, of the Anglophone countries (e.g. Reli Assets Holding Company or RAHCO in Tanzania)36.

35 In some countries, there was an independent Government Inspector of Railways but these were all too often seconded from the railway and with an ambition to subsequently return; this is not conducive to frank and fearless inspection of incidents for which your future superior is ultimately responsible. 36 But in some countries this function has been left to the ministries of transport.

39 RAILWAYS IN SSA

These organizations have varying powers but generally speaking should be the ‘counterparties’ to the concessionaires i.e. they are the ultimate owners of the assets, are responsible for ensuring they are maintained properly and for capital expenditure that is not the concessionaires responsibility. They are often also effectively the safety regulator, as they are the body that awards the operating licence on the basis of technical competence. The new regulatory bodies are often funded (at least in theory) from concession fees but this is an on- going problem in some countries and severely handicaps the ability of these bodies to provide effective regulation. Where there is strong modal competition, there is little need for sector-specific economic regulation, and this can normally be done through the general powers of a competition commission or suchlike37. In at least one case, however, a sector-specific authority (SUMATRA in Tanzania) has been established which regulates38 all land-based and marine transport and, in Senegal and Mali, railway regulatory agencies were created which were subsequently converted into a common railway monitoring agency for both countries. Many existing regulatory and legal frameworks stipulate that applicable tariffs for the transport of passengers and their luggage are freely determined. However, railways must communicate to the supervisory authority, usually the Ministry of Transport, any change in tariffs and obtain its agreement before implementation. The result is usually strong state influence over passenger tariffs, particularly in state-owned companies. It is not unusual for governments to take several years to approve an increase in passenger tariffs proposed by railway companies and, when increases are finally approved, they may not cover cost increases, particularly when inflation is high. 5.3 Governance and management of State-owned railways

Most of the remaining state-owned railways are subject to significant political and governmental influence. Specific arrangements vary across countries but the sectoral ministry (normally transport) exercises political and administrative control, while the Ministry of Finance exercises financial control. Board directors are generally a combination of ministry officials and internal senior management, themselves often appointed by the government, with occasional staff representation39. Oversight is nominally assigned to parliament but all too often parliamentary control is limited to an audit of company accounts presented in its annual report, often several years in arrears. Short parliamentary sessions also hinder the detailed review that effective control would require. Although the governing regulatory frameworks of the state-owned railways generally provide financial and management autonomy, and management methods are similar to those of private business, in practice they are considerably limited by the many opportunities for intervention by the state allowed under the legal and regulatory frameworks, at both the institutional and jurisdictional levels. This conflict between control and decision functions, and the frequent review by political authorities of initiatives and decisions taken by government’s authorized representatives in the corporation, does much to discourage management effectiveness. This is aggravated by the multiple objectives of governments and politicians, motivated by often mutually exclusive social, electoral and economic interests, which complicate both the management of state-owned companies and the evaluation of their performance. In executing their imposed public interest missions, state-owned companies’ management strategies often

37 E.g. in Zambia, the Competition and Fair Trading Act (CFT Act) as administered by the Zambia Competition Commission (ZCC) has broad powers of referral for the abuse of market power by dominant suppliers. 38 In this case, SUMATRA is also the safety regulator for both railways in Tanzania, whilst RAHCO only currently deals with the concessioned ex-TRC system. 39 This is not always the case, however; the Board of Botswana Railways generally includes at least three members from the private sector, including the Chairman.

40 RAILWAYS IN SSA focus on merely breaking even on a cash basis, leading often to financial difficulties when asset renewals fall due. 5.4 Structure of concessions

There are several different models for railway concessions but the most common are ones in which the state remains the owner of some or all of the assets (normally the infrastructure) but transfers the responsibility for operating the railway to a concessionaire under terms and conditions stipulated in a concession agreement. In particular, the concessionaire operates the railway as a business activity at its risk, cost and expense. The contractual arrangements range from a lease contract where the private operator only assumes the risks to operation’s revenue and costs, but not the risks to investments (as in Sitarail), on the one side to a traditional concession, where the private operator assumes all risks and where some or all of the assets revert to the state at the end of the concession period. The first railways to be concessioned were in West Africa, with the Abidjan-Ouagadougou railway linking Cote d’Ivoire and Burkina Faso being the first, followed at the end of the 1990s by Cameroon, Gabon and Malawi. The reform momentum accelerated in the 2000, as several countries felt the need to introduce similar reforms in their railway companies. Overall, since 1993, fourteen concessions and one management contract have been awarded in Africa, with a further four in process and others planned (Annex 1 and Figure 5.1).

Figure 5.1 Railway Concessions Awarded in Africa Since 1990.

Sizarail Transrail DRC Senegal/Mali 1995 – 1997 2003 2008

RVR Sitarail Kenya/Uganda Ivory Coast/Burkina 2006 Faso 1995

Canac/WACEM TRC Togo Tanzania 1995/2002 2007 Camrail Transgabonaise CEAR Cameroon Malawi 1999 Gabon 1999 1999 Madarail Railway operated by state RSZ Madagascar railway company Zambia 2003 PSP projects planned or 2003 underway Part of rail network now CDN under private management BBR Zimbabwe Ressano Garcia Mozambique Railway now under private Mozambique 2005 management 1997 Cancelled CCFB Mozambique 2005

41 RAILWAYS IN SSA

This restructuring has been a slow process for many countries, typically three to five years, and sometimes much longer:  The Sitarail concession was completed following three years of negotiations between the two states (Côte d’Ivoire and Burkina Faso) and a private consortium;  In Cameroon, the original process of concessioning took three years but the contract concerning passenger services was subsequently renegotiated four years later;  The signing of a concession contract (in 2005) for the Transgabonais (Gabon) took six years. The government had previously signed a concession contract with forestry companies in 1999 and a lease contract with a local mining company.  Preparation for concessioning in Tanzania began in 1997 and a concession was finally signed in 2007.  Ghana began the concession process in 2002 and it is still not well-advanced Due to their relatively small networks and traffic volumes, there is little scope on most African networks for on-rail competition and few governments have seriously considered the European model of full vertical separation. There are, however, some examples of third-party operators running over government lines: the Magadi soda works ran its own trains to Mombasa over the KRC line in Kenya (and still does, over the concessioned RVR) and TransAfrica ran a through container service for some years from Johannesburg to Mwanza 40 on Lake Victoria, including a gauge transfer at Kidatu. The Transrail concession excluded the Dakar suburban service (Box 2) and the SEFICS traffic, which both pay track charges to Transrail. In the case of Zambia, where there were extensive inter-mine operations in the Copper Belt, the concession offered some or all of three options: the mainline operations, the intermine operations and the passenger service. In the event the selected bidder wished to include all three, although he subsequently cut back substantially on the inter-mine services. The concessions do not always include the entire network. In some cases, branch lines were excluded, e.g. the Mulobezi and Njanji branches in Zambia, the Lumbo branch near Nacala in Mozambique and the St. Louis branch in Senegal. Table 5.1 summarizes the key features of the concession contracts to date. Their initial duration varied from 15 to 30 years, although this has been subject to renegotiation in some cases. The concessionaire is free to operate its activity as a business, with tariffs generally determined by supply and demand, although in some cases certain tariffs (e.g. passenger fares) are subject to some form of indexation.

40 It also hauled cement traffic, using only the TRC network, by arrangement for a few years.

42 RAILWAYS IN SSA

Table 5.1 – Key features of concessions 1993-2008 Initial capital Planned 5- Public Effective Initial of Concessionai year Service Countries date of duration of concession re investment Obligation contract contract (US $ (US$ million) (PSO) million) Western Africa Burkina Faso Yes but Sitarail 1995 15 8.8 63.3 Côte d’Ivoire renegotiated Mali Transrail 2003 25 17.2 55.4 No Senegal Canac 1995 n.a. Togo n.a. n.a. No WACEM 2002 25 Central Africa Cameroon Camrail 1999 20 18.5 89.6 Yes Sizarail 1995 n.a. n.a. n.a. n.a. DRC Vecturis(2) 2008 2 n.a n.a n.a Transgabonais 1999 20 n.a. n.a. n.a. Gabon e 2005 30 n.a. 157 Yes Setrag Eastern Africa Kenya RVRC 2006 25 n.a. About 100 Yes Uganda Tanzania TRC 2007 25 n.a. 88 Yes Southern Africa Malawi CEAR 1999(3) 20 n.a. n.a. Yes Mozambique CCFB 2004 25 19.7 152.5 Yes Mozambique CDN 2005 15 n.a. n.a. Yes Madagascar Madarail 2003 25 5 36.1 Yes Zambia RSZ 2002 20 6.1 14.8 Yes Zimbabwe NLPI 1998 30 n.a. 85(1) No Sources: Data collected from companies, Pozzo di Borgo (April 2006) (1) Reported construction cost (2) Management contract (3) Transfer finally occurred in 2005 Formal regulatory structures with real power are far from universal in SSA and many SSA rail concessions are thus potentially open to market abuse even though the concession agreements generally include protections; the two most common are the power to refer rail tariffs to either the Government or an independent authority and the power to allow third-party operators onto the railway. In practice, these protections are likely to be used very infrequently. The level of market power that most SSA concessions can exercise is generally limited by road competition along most or all of the routes they serve at tariffs which are generally competitive with rail (and make rail in many cases a price-taker). Whilst freight rates have increased in some cases following concessioning, the rail market share in general, unless it is an administered monopoly such as fuel, is not sufficiently large for such changes to significantly affect the selling price of commodities. In most cases, any such price rises have generally been accompanied by an improvement in service level, balancing out in terms of total cost to the customer. There are relatively few such cases, apart from bulk minerals such as manganese on the Transgabonais, where there are monopolies for freight41; in most cases the opposite is the case and freight rates are effectively set by the competition, either in terms of road or in terms of an alternative port for

41 And even here there is the (long-term) threat of the alternative rail route through Congo Brazzaville.

43 RAILWAYS IN SSA those routes where rail has a monopoly (such as Beira in the case of the Nacala corridor and Abidjan, Lome and adjacent ports in the case of Transrail). Many contracts also include third-party access clauses (as previously existed for SEFICS in Senegal and Magadi in Kenya) whereby third parties can run their own services on the concessioned infrastructure if they wish. The Camrail and Sitarail concession contracts contain track usage exclusivity periods of five and seven years, respectively, during which no other rail operator can operate trains on their networks but others, such as Madarail or Transrail, allow access from the start. The general absence42 of market abuse in practice was confirmed by an IBRD study in 200643, which examined in detail four transport corridors involving a railway concession (and in some cases an associated port) and could not find any evidence of abuse of monopoly. Only one contract out of the four (Sitarail) did not incorporate a clause that explicitly defines what constitutes evidence of undue market/pricing power, with the other three defining it in terms of: • when the tariffs applied by the operator are over twice as large as the costs incurred (including the depreciation and capital costs associated with the operation of the rolling stock); • when an operator openly discriminates against a client in terms of the transport conditions offered; or • when an operator refuses to provide services to a client. However, the first test in particular is only useful if a regulator has access to the relevant information, which is not always the case, and the skills to be able to undertake its own analysis. Even in the best of circumstances, information asymmetry between concessionaires and government authorities creates endemic oversight problems and these are multiplied when concessionaires actively refuse to co-operate, as has happened in at least one case in other matters. All the analyzed contracts stipulate that the Concessionaire is free to set tariffs for services other than public services, thus distinguishing between the generally deregulated freight tariffs and the passenger tariffs that tend to be controlled by the State. Regulated passenger services are often managed using pre-agreed schemes under which operators are eligible for financial compensation (the ‘Public Service obligations’ in Table 5.1) whenever the regulated tariffs do not cover their operating costs. However, in practice these schemes have often failed to protect R

e

s private operators when Governments have not honored their subsidy commitments. Passenger u

l

t

s

tariff indexation, where relevant (e.g. Sitarail, CEAR, and Transrail), is triggered by changes in o

f

conventional indices such as inflation. R

a

i

l

w

All four concession contracts also stop rail operators from using ‘promotional’ tariffs for more a

y

than a year if these tariffs do not cover their operating costs. The definition of ‘operating costs’ in P

r

i

v this context is open to considerable interpretation (many pre-concession tariffs could fail this test a

t

i

z on most SSA railways), its enforceability seems doubtful and its necessity is debatable. a

t

i

o

n

i

Where a concessionaire fails to comply with the terms of the concession, whether by design or n

A

f

by force of circumstances, through there are normally procedures by which the concession can r

i

c

a

4 42 The only examples encountered are anecdotal reports that the Comilog mineral trains on the Transgabonais systematically received priority over passenger services. This is not so much a concessioning issue as an access regime issue; it is an inherent risk that occurs when one operator has control over another’s services, as is effectively the case at present in Gabon, and can only properly be resolved by having a regulator with some teeth to monitor such situations. A second example arose in Zambia, where reportedly the concessionaire gives priority to longer-distance traffic from the Copperbelt to South Africa at the expense of local inter-mine movement of bulk minerals and feeder traffic to Tazara. This presumably reflects a deliberate decision by a concessionaire who is short of operating rollingstock, to concentrate his assets on those traffics which will be most profitable for him. In this case, the mines have the alternative of road transpor but, if the right rate could be established, presumably could also make a case for ZRS to invest in additional rollingstock to handle their traffic. 43 Pozzo di Borgo P. and al. (2006), Review of Selected Railway Concessions in Sub-Saharan Africa, World Bank Report.

44 RAILWAYS IN SSA be terminated. These have been rarely been applied to date, with only two44 concessions (Ressano Garcia, which never became operational, and Transgabonais) having been terminated and with two concessions (Transrail and Rift Valley) changing the operator. If a Government wishes to terminate a concession through no fault of the concessionaire, this is also normally covered by defined financial penalties which provide some protection to the concessionaire by requiring the Concessioning Authority to take over and/or service the debt related to any investment it has made and, in some cases, also making it liable for the projected benefits that each concession could generate for the remainder of the contract as well as making one-time payments. These conditions are substantial in many cases but would not be sufficient to deter a determined Government, although there could be substantial indirect costs in terms of the impact on any IFIs which participated in the concession financing and the private investor community in general. 5.5 Concessionaires

The rail concessions in SSA tend to attract a limited, pool of mainly foreign and South African private operators. These operators fall into two distinct groups: those which seek vertical integration of the distribution chain by acquiring dominant positions in specific production and transport sectors, and those which specialize in a single transport activity (e.g., railways or ports). The first group appears to accept low rates of return from individual elements of the distribution chain activities they operate (especially railways) as long as their overall control yields sufficient overall benefits. The main example of this type of operator has been the Bolloré group, which was the lead or the second largest shareholder in several SSA railway and port concessions, and which also has freight forwarder and (previously) agricultural production subsidiaries. The second category of operators, among whom the most prominent were Comazar (now defunct) and Sheltam from South Africa, NLPI from Mauritius and RITES from India, focuses on investment in transport operations, suggesting that these operations can be sufficiently profitable to attract specialist private transport operators. However, the business cases for these rail investments often appear weak45, suggesting that the companies that go after these concessions are sometimes focused on the financial benefits that can be extracted from managing large investment plans (financed for the most part from Governments) rather than long-term business cash flows. Private companies are the majority shareholders of all concessions to date (Table 5.2). State participation is highest in Mozambique, which holds 49 percent of both CCFB and CDN and is also a significant shareholder in the adjacent CEAR concession. In Madagascar, the government holds 25 percent of Madarail while governments own 10-20 percent in Sitarail, Transrail and Camrail. Local private participation in concessions has generally been relatively limited, and often fraught with problems during the bidding process; the highest level is in Madarail with 24% with 10-20 percent in Sitarail, Setrag and Camrail. Employee shareholding remains very weak, under 5 percent where it exists at all. Board membership of concessioned companies generally reflects the shareholding but in some concessions there are also government-appointed members.

44 Other than the special case of Sizarail in DRC, following a military coup. 45 One of the bidders for the Beira concession, awarded to RITES in 2004, was a Chinese consortium whose financial proposals had a calculated return on equity of only 2%.

45 RAILWAYS IN SSA

Table 5.2 – Initial concession shareholdings(1)

Concessionaire Shareholding Sitarail SOFIB46 67% ; Côte d’Ivoire and Burkina Faso governments 30% ; Employees 3% Transrail Canac-Getma (Franco-Canadian) 78% ; Mali and Senegal governments 22% Camrail SCCF47 (Cameroon) 85%; Cameroon government 10% Employees : 5% Setrag Comilog (France) 84% ; Local private operators :16% RVRC Sheltam (South Africa) 61%; other foreign 15%; local investors 25% TRC Rites 51%, Tanzania government 49% Edlow Resources/ Railroad Development Corporation (US) 51%; Mozambican investors CEAR (including CFM) 49% CCFB Ircon48 25 % ; Rites 26 % ; Mozambique government (through CFM) 49% CDN As for CEAR Madarail49: 51 % ; Madagascar government 25 % ; Manohisoa Financière 12.5 % ; Other private Madarail operators : 11.5 % RSZ NLPI (majority) ; Transnet (South Africa) (minority) (1) Ownership has since changed ion a number of concessions

46 Majority-controlled by Bolloré (France); 16% of this was intended to be sold on the Abidjan Stock Exchange 47 Société Camerounaise des Chemins de Fer. a holding company controlled by Bolloré but in which Comazar, a privately operated and managed company that included South Africa’s Spoornet and Transurb Consult, a subsidiary of the Belgian National Railways, had a substantial shareholding. Comazar has since sold its interests, in Cameroon and elsewhere, to a Belgian firm named Vecturis, set up by two ex-Comazar employees. 48 Indian Railways Construction Corporation 49 Majority-owned by Comazar

46 RAILWAYS IN SSA

6 OPERATIONAL PERFORMANCE

6.1 Overview

Given the poor condition of the infrastructure and rollingstock, low traffic levels and government ownership of most SSA networks, both labor and asset productivity (locomotive and wagon utilization) is low compared to railways elsewhere. However, one of the most visible characteristics of the concessions has been the sharp improvement in these indicators, partly because of growth in traffic but mostly due to major reductions in the workforces. 6.2 Labor productivity

Almost all railway companies have streamlined their work forces over the past 10-15 years, often as the prelude to concessioning but in some cases also as a general policy to improve efficiency. Nevertheless, labor productivity on most SSA systems is relatively low50 by world standards (Figure 6.1), with few railways achieving over 500,000 traffic units per staff p.a. Figure 6.1 Labor productivity on SSA railways (average 1995-2005)

3200 STATE - OWNED CONCESSIONED ) p 2800 m e / U

T 2400

0 0 0 (

2000 y t i v i

t 1600 c u d

o 1200 r p

r

u 800 o b a L 400

0 S N S T B S D C C G E B D N F G K T C S Z T U B N M M a C a a r p u o w t e e i e a R F o R i a g a a h a a t a h g z n m a d n i o t n m n m a c M E b a l d a r C C i n s e a a o z r a i a o y a g n o n a z n b w a s r i r w

r a p a r b n l o i d a n a i ( ( r i a r a i a n l a a n S C i l a i a i a a i a a e l n i i i n N F a l l t a d M C C K ) )

Average 1995-2005 Current

(1) No data available for FCE. Some railways in the ‘concessioned’ group were only concessioned close to or after 2005; this figure should therefore not be used to compare the impact of concessioning on productivity. ‘Current’ figures are for latest available year, typically 2005. Spoornet is in a league of its own, with an average productivity over the period of 2.5 million traffic units per employee. However, this average is boosted by the intrinsic high productivity of mineral-oriented lines: in 2005, when the average labor productivity of Spoornet was 3.3 million traffic units per employee, the productivity of the dedicated iron ore and coal export lines (Orex

50 Comparisons of productivity between railways need to be done with some care, as the staff employed in any individual system will be a function of how much work (especially major asset maintenance) is out-sourced and how much is done in-house. Measures of the work done by a railway are also difficult; the measure almost universally used is traffic units (the sum of net tonne-kilometers and passenger-kilometer) but, as a passenger-kilometer generally requires more resources than a net tonne-kilometer, this favours railways which are predominantly freight and particularly those which have a high proportion of minerals.

47 RAILWAYS IN SSA and Coalex) were 9 million and 38 million respectively, while that for the residual general freight business (about 40 percent of Spoornet’s traffic) was only about 1.5 million. The next highest average labour productivity over the period is Gabon, followed by Swaziland and Botswana. The Gabon productivity is helped by a high proportion of mineral traffic, for which moreover the trains are owned and operated by a third-party. Swaziland and, to a lesser extent Botswana, both carried substantial proportions of transit traffic during the period (75 percent in the case of Swaziland), for which the wagons are owned and maintained by third parties and which is relatively simple to operate. All three of these railways are either relatively new in railway terms or (in the case of Botswana) have received substantial investment in the last thirty years which is not yet life-expired. With the exception of these railways, productivity on most SSA railways is low, and in many cases very low. In North Africa, the railways in Morocco (ONCFM) and Tunisia (SNCFT) had labour productivities in 2007 of 1,160,000 and 749,000 traffic units per employee respectively, well above all State-owned SSA railways other than the Southern Africa trio of Spoornet, Botswana and Swaziland. The Algerian railway (SNTF), however, was similar to the typical SSA railway with a labour productivity of only 224,000 traffic units per employee. Following concessioning, labour productivities have generally increased as surplus labour disappeared. Most of the concessioned railways show a substantial improvement during the period; the main exceptions being those which were concessioned at the start. In some cases, the low-medium labor productivity reflects the continuing use of labor-intensive methods with relatively little out-sourcing (in 1958 the labor productivity in the UK railway industry was only 117,000 traffic units per employee) but in those railways with very low productivity it is the consequence of traffic declining without any adjustment to staff levels. When wages are very low, the direct financial impact is not always catastrophic but a large body of semi-employed staff has a corrosive effect on morale and is a strong disincentive for those who have the opportunity to improve efficiency. An important side-effect is that low-wage railways find it hard to recruit and retain technically competent staff, or to introduce the technology required to improve service levels, for which a better-paid and more skilled workforce is essential. 6.3 Rollingstock productivity

Although rollingstock productivity is commonly presented in terms of passenger-kilometers and tonne-kilometers per rollingstock unit within the fleet, this represents the combined effect of five factors, some of which are under management’s control and some of which are not:  The proportion of the fleet which is potentially useable  The proportion of the useable fleet which is available for use on any given day (this is normally termed ‘availability’)  The extent to which the available fleet is actually used (this is normally termed ‘utilisation’ and is usually measured in hours per day)  The commercial speed of the services operated by the railway (this determines utilization in terms of vehicle-kilometers)  The power of the locomotives and the size and loading of the carriages and wagons (this determines utilization in terms of passenger-kilometers ort net tonne-kilometers). The low utilizations reported by many SSA railways are a product of problems in all five of these areas. Many fleets include vehicles which exist on paper only; they may have been involved in accidents and never repaired, or have become obsolete in the case of specialized wagons, or have been a specialized type of locomotive which have been progressively cannibalized to provide otherwise unobtainable spare parts. Even though many such vehicles

48 RAILWAYS IN SSA may exist only in the form of a makers’ plate, it is often difficult to write them off in a State- owned railway and they thus remain in the nominal fleet until their accounting life has expired. Availability reflects the ability of the railway to maintain its useable rollingstock. This is influenced by how reliable the rollingstock is in the first place, the competence of the maintenance staff and the ability of the railway to ensure a supply of spare parts. Many SSA railways have a disparate collection of rollingstock of various makes and models. Some of these are relatively reliable but others are not, especially as they age (e.g. diesel locomotives over 30 years old). Maintenance problems are then often compounded by the lack of spare parts. In the case of locomotives these are almost all imported and thus require foreign exchange; the combination of foreign exchange procedures and a long and slow supply chain can often lead to locomotives waiting for a year or more for a spare part. When the locomotive is old and production has been discontinued, spare parts are often almost unobtainable and the railway then has no choice but to resort to cannibalism. Utilization of the available fleet in terms of hours is often very high, particularly of locomotives, as shortage of rollingstock will often mean intensive use of what is available. However, utilization in terms of distance traveled is a function of the commercial speed (i.e. the average speed of travel including the time spent waiting at intermediate stations en route) and commercial speeds on most SSA railways are relatively low. Where a railway has surplus assets (normally this is particular subfleets of carriages and wagons), utilization may also be low when there is no demand for a particular type of rollingstock. Finally, rollingstock productivity in terms of traffic units is heavily influenced by the power of the locomotives and the average load in the carriages and wagons. A 1000 HP locomotive will only ever be able to achieve half of what a 2000 HP locomotive can. Carriages on many of the remaining passenger services in SSA are often heavily loaded, in spite of low axle-loads but a wagon on a railway such as Tanzania will generally carry a maximum of 30 tonnes compared to the 55 tonnes that can be carried on a railway with a 20-tonne axleload. The combination of these factors means that aggregate rollingstock productivity on most SSA railways will always be low by world standards; better management can certainly improve productivity by disposing of surplus assets and improving the supply chain for spare parts, as several concessions have shown, but it can do relatively little to overcome the basic problems of low axleload and low commercial speeds. Locomotive productivity Locomotive productivity is generally low in SSA railways (Figure 6.2), with Swaziland being a long way ahead of the remainder51. The fundamental problem is fleet availability, with many railways having figures of under 40 percent. In the case of SNCC, which has fleets of electric and diesel locomotives, in 2008 only 10 of the 22 mainline electric locomotives and 13 of the 47 diesel locomotives were available on average, equivalent to rates of 45 percent and 28 percent respectively. Table 6.1 presents detailed availability statistics for a better-performing railway, Tanzania in 2001 prior to concessioning, to demonstrate the impact on utilization statistics of ‘dead’ locos and the level of availability. Out of a total fleet of 121 locomotives, only 96 were active; the remainder were either under long-term repair or were not capable of being repaired. Of the active fleet, 74 percent were available on any given day, with this proportion ranging from over 80 percent for some of the mainline locomotives to 50 percent for some of the shunting locomotives. However, when the non-active fleet is taken into account, availability drops from 74 percent to 59 percent.

51 However, Swaziland is essentially a transit railway and it uses Spoornet locos (which may or may not be included in the reported fleet) on many of its operations.

49 RAILWAYS IN SSA

Figure 6.2 Locomotive productivity (average 1995-2005)

100

90 STATE - OWNED CONCESSIONED ) o

c 80 o l / U

T 70

n o i l

l 60 i m (

y 50 t i v i t

c 40 u d o

r 30 p

o

c 20 o L 10

0 S N S T B S D C C G E B D N F G K T C S Z T U B N M M C a a a r p u o w t e e i e a R F o R i a g a h a a a t a h g z n m a d n n i o t m n m a c M E a b l d a r C C n s i e a a z r o a i y a o a g n n o z a b n w a s r i r w

r a a p r b n l o d i i a n a ( ( r i a r a i a n l a a n S C a i l i a a i a i a a e l n i i n i N F a l l t a d M C C K ) )

(1) No data available for FCE and Transrail. Some railways in the ‘concessioned’ group were only concessioned close to or after 2005; this figure should therefore not be used to compare the impact of concessioning on locomotive productivity. The average distance traveled per day by locomotives that are in use (excluding the shunting fleet) is around 291 km; as the average commercial speed at the time was around 18 km/hr, this implies each loco operated for around 16 hours per day.

50 RAILWAYS IN SSA

Table 6.1 Locomotive utilization and availability – Tanzania 2001 Number of locomotives Km/loco/day Class Type Age Availabl Availabl Total Active Active Total e e Diesel electric 36 Shunter 22 20 16 8 - - - 73 Mainline 26 15 10 8 210 168 112 87 Mainline 34 7 2 1 123 53 15 88 Mainline 26 35 31 26 344 289 256 89 Mainline 9 9 8 5 362 232 206 Subtotal 86 67 48 315 247 191 Diesel hydraulic 35 Shunter 29 4 2 1 - - - 37 Shunter 16 5 5 4 - - - Branchlin 64 22 21 17 13 235 183 148 e Branchlin 65 9 4 4 3 231 192 192 e Subtotal 34 28 22 234 185 155 29 Steam 46 1 1 1 - - - Total 121 96 71 291 229 181 Availability 59% 74%

The locomotives that are working are thus used relatively intensively but overall productivity is low, below 100,000 km p.a. for the active non-shunting fleet and below 70,000 km for the total fleet, because of the low commercial speed. Table 6.1also shows a final trap in assessing locomotive productivity. The inclusion or exclusion of shunting locomotives, around 25 percent of the fleet in this case, will have a significant effect on any indicators and any comparison needs to be clear about whether these have been included or excluded. Carriage and wagon productivity Carriage productivity shows a very wide range, from around an average over the period of 5 million passenger-km/car annually for Nacala, Camrail, Ethiopia and Tazara to under 500,000 passenger-km/car in DRC and Sudan. Some of the high productivities reflect very small fleets of carriages which are very overcrowded when services are operated. More useful figures are from railways such as Camrail, which operate a regular service with reasonable load factors; for such a railway, the productivity is typically the product of an annual distance traveled per vehicle of 100,000 – 130,000 km and an average load of 40 - 50 passengers. Availability is generally reasonable if there is sufficient demand to keep most of the fleet operating, often well over 80 percent.

51 RAILWAYS IN SSA

Figure 6.3 Passenger car and freight wagon productivity (average 1995-2005)

5500 1500 5000 STATE - OWNED CONCESSIONED STATE - OWNED CONCESSIONED ) ) r n a 4500 o c g / a m w k 4000 / p k

t 0

n 1000

0 3500 0 0 ( 0

0 y ( t 3000 i y v t i i t v c 2500 i t u c d u o d r 2000 o p

500 r e p

g 1500 n a i o r g r a a 1000 C W 500

0 0 S N S T B S D C C G E B D N F G K T C S Z T U B N M M S N S T B S D C C G E B D N F G K T C S Z T U B N M M a C a a r a C a a r p u o w t e e i e p u o w t e e i e F o a F o a a R R i g a a R R i g a h a h a a a a a t t h a h a g g z n a m z n a m n n n n o d t i o d t i n m n m m a c m a c M a b M a b E E l d l d a a C C r C C r s i e n s i e n a a a a o z r o z r i y i y o a a a o a a a g n g n n o n o z a z a n b n b w w r a s r a s r r i i w w

a a r r p a p a r b n l r b n l o i d o i d a n a a n a i ( ( i ( ( r r a i r a i r a a a i a i n l n l a a n a n a S C S C i l a i l a i a i i a i a a a a i i a a a a e l e l n n i i i i i i n N n N F F a a l l l l t t a a d d M M C C C C K K ) ) ) ) Passenger carriages Freight wagons (1) No data available for FCE and Transrail. Some railways in the ‘concessioned’ group were only concessioned close to or after 2005; this figure should therefore not be used to compare the impact of concessioning on rollingstock productivity. Wagon productivity is around 1 million ntk/wagon in Swaziland (but helped by the transit traffic carried in ‘foreign’ wagons), with productivities of over 500,000 ntk/wagon on several other systems. However, on many State-owned railways (including some only concessioned at the end of the period) wagon productivity is very low, below 200,000 ntk/wagon; this is the product of demand reducing while the wagon fleet (which can easily have a life of 50 years or more) remains unchanged. On a busy railway, wagon availability should be well over 80 percent, as maintenance is straightforward and normally only a limited range minimum of spare parts is required. A wagon should be able to travel at least 50,000 km p.a. on a typical railway52 but this depends critically on ‘cycle time’, the interval between successive loadings, and this in turn is primarily a function of the efficiency of loading and unloading on the part of both the railway and the customer. The average load in an SSA wagon (taking account of empty running) is about 20 tonnes on the 15 tonne axle-load systems and about 30 tonnes on the 18-tonne systems. A well-run railway should thus be able to easily achieve 1 million ntkm/wagon p.a.; the low productivities on many of the railways over the past decade reflect very low annual utilizations, around 10,000 km or less, which in turn is a product of wagon fleets with many surplus (and generally obsolete) wagons 6.4 Impact of concessioning on productivity

Both labor and asset productivity is higher in concessioned companies than state-controlled companies in SSA. Labor productivity of concessioned companies is on average twice as high as in public companies.

52 And much more if the train is a dedicated train carrying a single commodity between two locations. In the early 1980’s, Zimbabwe railways ran such a ‘block train’ carrying coal between Hwange and the Zisco steel plant at Kwekwe , with the wagons traveling around 200,000 km p.a.

52 RAILWAYS IN SSA

Figure 6.4 Impact of concessioning on labor and asset productivity)

700 60

600

) ) 50 f o f c a o t

Date of l / s /

500 concessioning U U T

T

40 n 0 o 0 i l 0 l ( i

400 y m t ( i

v

y 30 i t t i c v i

u 300 t d Rivi Rivi bridge c o u r d

p collapse 20 o r r p

u 200

o o b c a o L

L 10 100 Civil war in Ivory Coast from late- 0 2003 0 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 Years since concessioning Years since concessioning

Sitarail Camrail CEAR RSZ Sitarail Camrail CEAR RSZ Labor productivity Locomotive productivity

7000 1200 Sitarail out-sources

passenger services

)

) 6000 n r

o 1000 a g c / a w m /

k 5000 m p

k 800 t 0 0 n

0 0 (

4000 0 y 0 t ( i

v y i 600 t t i c v i u

3000 t d c o u r d p

o 400 r e p g 2000

a i n r o r g a a

C 200 1000 W

0 0 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 Years since concessioning Years since concessioning

Sitarail Camrail CEAR RSZ Sitarail Camrail CEAR RSZ Carriage productivity Wagon productivity Year of concessioning = Year 0; index at Year 0 =100

Figure 6.4 summarizes four key indicators(i) traffic units (passenger-km plus net tonne-km) per staff, an indicator of labor productivity; (ii) traffic units per locomotive53; (iii) net tonne-km per wagon, and (iv) passenger-km per carriagefor each of the three for the five years prior to concessioning and the period since. As discussed above, all these indicators have weaknesses as they stand (for example, they take no account of loco or wagon capacity, nor of passenger car occupancy) and can be considerably improved with more detailed data. However, they are are sufficient to provide a broad picture. Labor productivity increased steadily in all the concessions. In the case of Sitarail, the service suspensions caused by the civil war in 2003-4 years interrupted the improving trend but it returned to its upward trend when a stable situation returned. Camrail labor productivity increased sharply on concessioning as traffic grew; it then stabilized but now appears to be increasing again. CEAR productivity grew sharply on concessioning when it only re- employed about two-thirds of the previous workforce whilst growing traffic by about 30 percent on an adjusted annual basis. Productivity in 2003-4 fell because of the disruption caused by Rivi Rivi bridge, which left the northern half of the network, and the associated staff, with very little traffic54. The takeover of the Nacala line in 2005, and the reopening of Rivi Rivi should give a substantial boost to the figures for subsequent years (currently not available). Similar figures are likely to come from most of the other recent concessions: Madarail reduced the workforce by 50 percent, both the Mozambique concessions by about

53 The CEAR locomotive productivities ignore the scrap locomotives that were purchased for spare parts. 54 These statistics also demonstrate one of the traps in such broad measures of productivity, as during 2004 the CEAR locomotives did 25 percent of their work on hire to CFM, which is not reflected in the traffic statistics.

53 RAILWAYS IN SSA

60 percent, the Zambian concession by over 30 percent and the Senegal concession by about 40 percent. Locomotive productivity increased sharply in Sitarail and Zambia, mostly due to scrapping surplus equipment. Productivity in Cameroon has increased steadily as traffic increased but a major jump (not shown on the graph) occurred in 2007 when new and more powerful locomotives were introduced, approximately halving the fleet size and thus doubling productivity. There has been little change in CEAR but in practice much of their fleet has not been used with some locos bought not to operate but to be a source of spare parts. Both the carriage and (especially) wagon fleets also show improvements in productivity following concessioning, again partly because of traffic growth and partly because of scrapping surplus stock. Although such workforce and rollingstock reductions feed directly through into the productivity figures, they paint a misleadingly gloomy picture of the middle managers of the previous government railways. In many cases key managers remained after concessioning and the high level of surplus labor in the system often reflected not so much management capability as political decisions made in the past to protect employment levels irrespective of railway efficiency. Neither is this a special characteristic of the rail industry; it is common in many parastatal organizations, both in Africa and elsewhere. Concessionaires, almost without exception, also operate their railways with better asset utilization. Asset productivity has also generally increased. In some cases this reflects greater use being made of assets that were previously lying idle (this is particularly the case with wagons). In others it reflects the previous bureaucratic difficulties in writing-off publicly-

R

owned assets that are surplus to requirements that were left behind by the concessionaire. e

s

u

l

Nevertheless, whichever asset class is examined, and whether infrastructure or rollingstock, t

s

o

it is achieving better productivity under concessioning than before. f

R

a

i

l 6.5 Service quality w

a

y

P

r

i

The quality of service offered by many SSA railways is poor. The key attributes of both v

a

t

i

passenger and freight services are adequate quantity (supply), quality (safety, security, z

a

t

i cleanliness, speed, reliability) and timeliness (availability). Few State-run systems score well in o

n

i

anything more than isolated cases, as discussed in earlier chapters. Concessioned railways n

A

cannot necessarily improve transit time but they do generally try and address other aspects of f

r

i

c service quality such as safety and security and reliability. a

2

Safety in particular is a major weakness of many railways. In the last ten years there have been 8 several major accidents on SSA railways, some due to basic operating failures, with major losses of life. There are also many derailments on most systems. While this is an occupational hazard on all railways, the frequency on many SSA railways is extremely high by any standard. Many SSA railways report over 100 derailments per year and some report over 200 or even 300. The definition of a derailment varies from railway to railway, and many often occur at slow speeds in yards and pose little, if any, threat to safety. Nevertheless, the whole of the US system typically only has around 2000 derailments each year, of which one-third is on mainlines as opposed to yards and sidings. Canada has around 150 mainline derailments each year and India less than 100. When related to the volume of traffic, SSA derailment rates are an order of magnitude greater than most other railways. Even controlling for track quality does not close the gap: there is a derailment on Class 2 track on US non-Class 1 railways (which has a 40 km/hr speed limit and is broadly equivalent to the poorer parts of the SSA mainline network) for every 10 million wagon-km; the equivalent rate for SSA railways appears to be some 30 times greater.

54 RAILWAYS IN SSA

7 FINANCIAL PERFORMANCE

This chapter first describes the revenue and revenue structure of African railways activities. It then reviews financial management and some key financial ratios. Finally, it discusses the financial structure of the main concessions and future investment needs. 7.1 Revenue and Cost Structure

Chapter 4 discussed freight and passenger traffic, the two main sources of revenue on almost all SSA railways. Most railways in addition earn revenue from minor operational activities e.g. several railways earn revenue from trackage (or access) charges levied on third parties running over their network and there is generally some revenue from sales of scrap, rentals, advertising and the like. However, these ancillary sources of income rarely generate more than about 5 percent of the total revenue of a railway. Some government railways receive annual payments from the government to enable them to breakeven on a cash basis and this is often included as ‘revenue’ in the published accounts. The cost structure of railways can be analyzed in two ways:  firstly by cost type (salary, materials, sundries (which normally includes maintenance undertaken by third parties and fuel). This is the standard presentation in most annual reports.  secondly, by functional area (maintenance, operations, administration). This is sometimes available where there has been some internal reorganization but normally this information is only available from the general ledger (and not always then). It is, however, the basis for any analysis of financial performance, especially the benchmarking of costs against other railways. Figure 7.1 gives the functional cost structure of three government railways in SSA around 2002, two (A and B) from low- income countries and one (C) from a medium-income country. Terminal costs represent about 10 percent of total costs; train operations costs (loco crew, guards and signalmen and safeworking staff) are about 15 percent and loco and vehicle maintenance about 20 percent. Fuel is about 20 percent on average, although it varies from railway to railway depending on terrain and general operating characteristics (a railway in flat terrain may only use 50 percent of the fuel used by one in hilly terrain). Infrastructure and signal maintenance is about 20 percent of the total and finally corporate administration and overheads are the remaining 15 percent55. This overall structure, which covers recurrent cash costs only, is not dissimilar to other railways. It excludes depreciation, which in many SSA railway accounts is virtually meaningless but which is typically about 10-15 percent on well-managed railways; it also excludes pension payments which on many railways are cash payments to past employees rather than a contribution to a funded scheme for current employees; these would typically add a further 10 percent to total costs. The precise cost structure on any given railway is a reflection of the traffic mix, operating conditions and management capability. It is also a function of asset condition and funding capacity; on many railways periodic maintenance (e.g. track renewal or the re-engining of locomotives) can only be done when funds are available; such expenditure is up to half the long-run maintenance cost, low maintenance expenditure is just as often a sign of deferred maintenance and asset deterioration as it is of skillful management.

55 Comparable figures for the Indian Railways in 2004/5 were: train operations (including terminal) 33 percent, loco and vehicle maintenance 18 percent; fuel 26 percent; infrastructure maintenance 17 percent and corporate administration 6 percent.

55 RAILWAYS IN SSA

Figure 7.1 Cost structure – selected systems – 2002

100%

90%

e 80% r u t i d 70% n e p x 60% e l a t o

t 50% f o

e 40% g a t n e 30% c r e

P 20%

10%

0% A B C Average

Passenger terminal Freight terminal Traffic Loco maintenance Vehicle maintenance Fuel Infrastructure Signals Corporate overheads

Figure 7.2 Unit costs – selected systems – 2002

14

12

10 ) $

S 8 U ( t s o c t i 6 n U

4

2

0 Loco mtce/loco-km Wagon mtce/ 00 Carriage mtce/ 00 Track mtce/track- Fuel/000 gtkm wagon-km car-km metre

A B C

56 RAILWAYS IN SSA

Unit costs56 for selected activities for the three railways are compared in Figure 7.2. In terms of the absolute level of expenditure, they are comparable with other railways world-wide (e.g. US $1.50 per loco-kilometer is a reasonable figure for a fleet of old locomotives, and even better for the small mixed fleets which are common on many SSA railways). However, the problem is generally what is achieved for this expenditure, which is generally only low levels of availability and utilization; this is not necessarily the fault of the individual engineers or mechanics but rather the consequence of old assets which in many cases are well past their economic life. The actual cost levels per unit of traffic are a combination of:  the unit costs for functional activities, which vary with managerial efficiency, the age and condition of assets and the extent to which maintenance which should be done is actually carried out  the mix and type of traffic, which determines how many resources are required to haul traffic e.g. how big trains are (which is also a function of the terrain and grades), whether passenger trains are mainline trains with, say, fifteen carriages or branchline trains with, say, four carriages.  the utilization and occupancy of services e.g. the extent of backloading, the carrying capacity of wagons (which is generally a function of the axleload but also varies from commodity to commodity) and the passenger load factors achieved. As an example, the cost of a passenger-kilometer in a sleeping car is clearly much greater than the cost of a passenger-kilometer in a crowded economy carriage. These factors need to be taken into account in interpreting Figure 7.3, which gives the average expenditure per unit of traffic for a range of SSA railways between 2000 and 200557. Figure 7.3 Unit costs per traffic unit– average in period 2000 – 2005

8

7

6 t i n

u 5

c i f f a r t

4 r e p

c

S 3 U

2

1

0 l l i i a a a a a a B N R N O i i M r a a y n n F r b r n B A D F a C n a a F a a z E C F m e C C m t h w i z a C a a C C K O s n G S T t Z C a o T B

56 i.e. the cost per unit of output, normally defined in terms of measures which have a direct causal relation such as loco-km for locomotive maintenance etc, as shown in Figure 7.2. 57 The figures are averages for the years for which data is available. The cost, excluding depreciation and concession fees for the concessioned railways where possible, is expressed per traffic unit i.e. the sum of net tonne-kilometers and passenger-kilometers

57 RAILWAYS IN SSA

Most railways lie in the range 2-5 c/traffic unit, with the lowest cost in Botswana, a relatively flat railway with modern equipment, good track and a base mineral traffic. The highest costs are in the low-volume systems of Benin and Congo Brazzaville [need to check Tazara]. 7.2 Financial results

Although almost all railways produce annual accounts, these are generally of little value other than as an official record of revenue earned and expenditure made. Cash shortfalls are generally made good by grants from governments (and often included as revenue) while depreciation is recorded on a number of bases but, as a non-cash item, is of little practical consequence for most railways. As a result, most railways just about breakeven on a cash basis, after receipt of government support, but with the proviso that there is almost always a substantial amount of maintenance which is deferred. When the maintenance backlog becomes too great, it is typically addressed using a loan with the expenditure being treated as investment. The two companies which have been concessioned the longest (Sitarail and Camrail) both make modest profits58. CEAR, which suffered long delays in finalizing the companion CDN concession in Mozambique, made working losses for several years. The performance of RSZ is unknown and it is too early to judge in the case of Tanzania and Kenya. One minor, but positive, factor is that concessionaires normally take a vigorous approach towards dealing with some of the dubious practices entrenched in some railways. Railways have traditionally provided many opportunities for employees, whether individually or in groups, to operate “businesses within businesses”; examples range from small-scale theft, such as stealing locomotive fuel or pocketing cash fares on passenger trains to larger-scale operations such as the systematic overlooking of demurrage or, particularly on railways which have an apparent price advantage over road, payments for wagon supply or expedited (i.e. non-delayed) transit. Staff have sometimes been allowed to continue these practices because of government controls on their salaries as public servants; attracting and keeping qualified staff then requires a supplementary remuneration mechanism but this often has a negative effect on rail operations, particularly those practices which attempt to ration wagon supply. 7.3 Passenger services

Passenger services are generally regarded as incorrigible loss-making activities. Whilst they generally do not contribute significantly to the cost of maintaining infrastructure or to covering corporate overheads, neither are they universal loss-makers on a marginal basis.

58 The financial performance of these two companies is discussed in detail in ‘ESW concessions etc’ PPdiB, World Bank 2006

58 RAILWAYS IN SSA

Figure 7.4 Revenue and cost per carriage-kilometer (selected systems) – 2002

1.60 1.60

1.40 1.40

1.20 1.20 m k m - 1.00 r k 1.00 - a r c a r c

0.80 e r

0.80 p e

p $

$ 0.60 S

U S 0.60 U 0.40 0.40 0.20 0.20 0.00 0.00 A B C A B C Above-rail (AR)/car-km Depreciation/car-km Access/car-km Revenue per carriage- kilometer Cost per carriage-kilometer Figure 7.4 shows absolute levels of revenue and cost per carriage-kilometer for the three SSA systems in Figure 7.1; this statistic is the basic driver of passenger operating costs and probably the best single measure of how successful the railway is at trading off price, capacity and demand59. The figure shows three components of cost:  the avoidable costs of train operation, i.e. if the costs which would be avoided if the some of the services ceased operation. These can conveniently60 be measured as the costs of train crew, rollingstock maintenance, fuel (or power) and passenger handling costs (ticket- selling/commission and some station staff).  rolling-stock capital renewal costs61, which can be derived by converting the capital cost into an equivalent annual sum, based on the asset life and using a real discount rate62. These costs are effectively sunk63 until replacement rollingstock is required.  an allocation of infrastructure operation and maintenance costs, for convenience termed ‘access charges’ in this report. Although many different procedures have been devised for determining how much should be charged to the different types of service, for simplicity a straightforward full-cost allocation method has been used. A final potential component (not shown in Figure 7.4) is the capital cost of infrastructure renewal in the cost base. This cost occurs very infrequently on most low-density railways (assuming routine maintenance has not been neglected). Only a few passenger services are ever able to make any contribution to this cost even though it is at least partly variable with usage64. Although the three railways have rather different operations, with average train sizes ranging from 9 to 16 vehicles, and with commercial65 speeds ranging from 28 km/hr to 50 km/hr, the average revenues and costs are relatively similar. Revenue per carriage-kilometer is the product of tariff and occupancy – similar results can be obtained from a high occupancy-low fare railway as from the reverse. Cost per carriage-kilometer is also a function of average speed and, particularly for depreciation, the annual utilization of rollingstock.

59 For passenger railways, as with airlines, operating costs are primarily determined by the number of services offered, whether or not there are passengers using them. Unit costs per seat-km or per carriage-km are therefore a more useful measure than costs per passenger-km, which reflect marketing decisions which have comparatively little impact on operating costs. 60 A proportion of infrastructure costs is also variable with usage and should strictly speaking be included. However, for many of the more basic low-density railways the incremental impact of passenger services on infrastructure is relatively small unless the services are withdrawn in their entirety, when some significant changes in track and signaling standards can often be made. 61 For convenience, these costs have been termed ‘Above-rail depreciation’ in the discussion that follows, assuming depreciation is based on the renewal cost of assets rather than the historic cost. Costs calculated in this way are unlikely to appear in the accounts of any railway in SSA. 62 Assumes a 4 percent real rate of return 63 Although there is an international market in used rollingstock , most of that currently in use in SSA outside South Africa is probably unsaleable 64 Even fewer are able to make a contribution to the initial capital cost of construction. 65 Commercial speed is the end-to-end time including stops at stations and for operational reasons.

59 RAILWAYS IN SSA

The rolling-stock renewal costs (termed ‘above-rail depreciation charges’) depend on the rollingstock assigned to passenger services on each system (excluding that which is obsolete and/or surplus) and estimated replacement costs. Unlike the above-rail working expenses and the allocated infrastructure costs, these costs are independent of labor costs and productivity; instead they primarily depend on the average utilization of the rollingstock. This in turn depends on the average train operating speed, the intensity of services (so there are not long delays between consecutive services) and the degree to which rollingstock is unserviceable because it is under repair in workshops and depots. Not surprisingly, therefore, railways with low operating speeds and low service frequencies (typical of most of SSA) have very high depreciation charges compared to operating costs. These can be dramatically reduced by improved utilization; an efficient railway be able to achieve a cost per kilometer of around 25-30 percent of the levels in Figure 7.4. However, on most SSA railways this requires capital expenditure to increase commercial speeds (and hence kilometers per hour operated) and more intensive services (and reduced maintenance downtime) to increase the hours rollingstock is used each year. Figure 7.5 combines the costs and revenues of Figure 7.4 to show three indicators of financial cost recovery:  The most basic indicator of financial performance compares revenue with the avoidable costs of train operation; if the revenue only just covers the expense of the day-to-day operation of the train it can clearly not make any contribution towards rollingstock capital or infrastructure costs.  The second indicator includes rolling-stock capital renewal costs in the cost base. If revenue covers this cost threshold then the service is able to cover replacement of its rollingstock when due, but will still require someone else (either the taxpayer, or freight traffic) to pay for infrastructure.  The third indicator includes an allocation of infrastructure operation and maintenance costs. This shows whether an operator would be able to operate commercially without any contribution from third parties for infrastructure.

60 RAILWAYS IN SSA

Figure 7.5 Passenger cost recovery – selected systems – 2002

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0% A B C

AR AR+depreciation AR+depreciation+access

None of the three railways is able to cover its above-rail working expenses. The only railway to come close is railway A, which has the highest earnings per carriage-km. The other two railways cover about 50% of their above-rail working expenses.

When rollingstock capital is included, even railway A performs poorly, based on current levels of rollingstock utilization. If depreciation charges could be reduced, either through assuming cheaper secondhand equipment or by increasing utilization, cost recovery could possibly be improved to 50 percent. When full access charges are included (but without infrastructure renewal), the cost recoveries fall to between 20 and 40 percent, which might improve to between 30 and 65 percent with improved utilization. Some of the tariffs (including those in railway A) are essentially administered and this tariff-setting is done within a framework which only includes a subset of the costs included in Figure 7.5. Nevertheless, many of the more poorly-performing systems in SSA would be unable to cover above-rail working expenses even if they had freedom to set their own tariffs, at least on a system-wide level.

In the early 2000s at least, long-distance passenger railways needed to earn around US$1 per carriage-kilometer to ensure a long-term future. Earnings of US$0.75 per carriage-kilometer would have been enough to cover recurrent operating costs, and do a reasonable amount of the periodic maintenance required for the rollingstock but all asset renewal costs (i.e. new locomotives and new carriages) would have had to be funded by grants from third parties. Most economy class coaches on SSA railways can carry about 80 passengers and a dynamic load factor of 70 percent is a realistic, if challenging, target, taking into account part-way travel, seasonality and so on. This then implies a minimum unit tariff for third-class travel of about US

61 RAILWAYS IN SSA

1.5 -2 cents/passenger-kilometer; at any level below this Government support would be essential66. 7.4 Freight services

In contrast to passenger services, freight services are normally capable of at least covering their avoidable operating costs and, with luck, earning enough to also cover infrastructure costs and even rollingstock capital costs. Figure 7.6 shows absolute levels, for the same three SSA systems analyzed in the preceding section, of revenue and cost per wagon-kilometer, the basic driver of freight operating costs, with cost being divided into the avoidable costs of train operation, rolling-stock capital renewal costs and an allocation of infrastructure operation and maintenance costs. Figure 7.6 Revenue and cost per wagon-kilometer (selected systems) – 2002

1.40 1.40

1.20 1.20

m 1.00

1.00 k m - k n - o n g

o 0.80 a

g 0.80 a w

r w

e

r 0.60 p e 0.60 p $

$ S 0.40 U S

U 0.40 0.20 0.20 0.00 0.00 A B C A B C Above-rail (AR)/car-km Depreciation/car-km Access/car-km Revenue per wagon- kilometer Cost per wagon-kilometer The cost range reflects the differences in train operations, with average train sizes ranging from 18 to 27 wagons, and with commercial speeds ranging from 18 km/hr to 37 km/hr (the larger trains and higher speeds giving the lowest costs). Revenue per wagon-kilometer is the product of tariff and average loading – and the latter is the product of wagon capacity in tonnes, the density of the freight and the percentage of empty running. As with passenger services, the rollingstock renewal costs primarily depend on the average utilization of the rollingstock, itself a function of the average train operating speed, the intensity of services (so there are not long delays between consecutive wagon loadings) and the degree to which rollingstock is unserviceable because it is under repair in workshops and depots; railways with low operating speeds and slow wagon turnround thus have higher depreciation charges. Improved utilization will reduce the ‘depreciation’ charges shown in Figure 7.6 and an efficient railway should be able to reduce the cost per wagon-kilometer to around 70 percent of the levels of Railway C, given some investment in infrastructure and maintenance facilities. Figure 7.7 combines the costs and revenues of Figure 7.6 to show the same three indicators of financial cost recovery as were presented for the passenger services.

66 Lower load factors (which are a function of service frequency and train size) would increase this minimum, although in most cases it would still be comparable with bus fares. The main difficulties facing rail remain service frequency and travel time.

62 RAILWAYS IN SSA

Figure 7.7 Freight cost recovery – selected systems – 2002

250%

200%

150%

100%

50%

0% A B C

AR AR+depreciation AR+depreciation+access

All three railways comfortably cover their above-rail working expenses. They more-or-less cover their above-rail and rollingstock charges; if depreciation charges can be reduced, either through assuming cheaper secondhand equipment or by improving utilization, cost recovery would probably be comfortably above 100 percent. However, when full access charges are included (but without infrastructure renewal), only Railway C has a cost recovery above 100 percent. The other two operations are thus not sustainable in the long-term unless depreciation charges can be reduced or the government contributes to the cost of infrastructure. Even the best-performing railway, Railway C, does not earn enough to cover the cost of renewing infrastructure. Freight railways at the date of the analysis (2002) therefore needed to earn around $0.80-1.00 per wagon-kilometer whilst keeping their operating costs to around $0.60-0.80 if they were to be self-sustaining67. This can be achieved but requires a combination of two-way loading and higher-yielding traffics. In turn, this requires strong marketing and a higher level of service to attract the higher-value traffic An order-of-magnitude estimate of the breakeven unit revenue required at that time can be made assuming that 40 percent of wagons were backloaded (this depends on the type of freight – specialized rollingstock carrying traffic like petroleum products or bulk cement has to return empty except in very unusual circumstances) and had a carrying capacity of 30 tonnes. Assuming also that backloaded tariffs were 60 percent of forward-loaded tariffs gives earnings of $0.93/wagon-kilometer from a forward rate of 5 c/net tonne-kilometer. The forward unit rate has clearly been achievable; the key is finding sufficient backloads and achieving good rollingstock utilization, both of which concessionaires are normally good at.

67 This analysis excludes concession fees, which would typically represent around $0.05 per wagon-kilometre.

63 RAILWAYS IN SSA

7.5 Concession financing issues

Two key features of many of the concession contracts are firstly, a substantial contribution from low-interest sovereign loans onlent to concessionaires on terms which are not available commercially68 and, secondly, relatively low proportions of equity being provided by concessionaires. Figure 7.8 Financing structure of selected concessions

180 100

160 90

80 140 t

70 b

120 e d

y

60 b n

o d i

l 100 e l i d m 50 n

u f S

80 s U t $ 40 e s s a

60

30 % 40 20

20 10

0 0 Madarail Sitarail Beira Camrail Transrail KRC URC Zambia TRC

Equity IFI/bilateral Other % debt

(1) The Madarail concession is shown as after the September 2005 restructuring, the main consequence of which was transform what was originally an IFI Euro 21 million loan onlent from the Government to Madarail into a grant; this has been included in the IFI/bilateral funds in the chart. (2) As the Sitarail concession is an affermage, Sitarail does not carry the debt contracted by the State holding companies (i.e. USD 56.7 out of 63.6 million). Nevertheless, since Sitarail services the debt, it has been included in the chart. All of the concessions planned to finance over 80% of their investments with debt 69, a proportion at the top end of what is normally deemed desirable for any financial venture. The share of the investments privately financed is in many cases well below 50%, and those concessions which planned a substantial contribution from commercial borrowings (Transrail, Kenya/Uganda and Zambia) have faced consistent criticism of their lack of investment in practice. The relatively low amount of equity put into the concessions in most cases is more than compensated by the rollingstock transferred to the concessionaire (even though much has been in poor condition) and this has resulted in a significant transfer of the financial risks associated

68 Transport concession financing in SSA is usually based on Government sovereign loans from IFIs which are on-lent to the concessionaires, the same practice as used in many other countries for loans to publicly-owned railways. In the mid 1990s, this was usually carried out at a premium (e.g. the Sitarail loan was an IDA loan to Côte d’Ivoire with an interest rate of 0.75% p.a. but it was on-lent to Sitarail at 8.00% p.a.). This premium has been reduced sharply in subsequent transactions to attract potential operators, sometimes to 0% as in the case of Madarail; as a result, the average interest of the Madarail operator’s debt is only 1.73% with a 7 year grace period and a 25 year tenor. 69 In practice, even this level of reported equity tended to be exaggerated as the Governments’ share usually involved in-kind contributions (e.g., rolling stock, spare parts or buildings) with a limited cash value.

64 RAILWAYS IN SSA

Box 7.1 Robustness of bid financial projections

In many cases, initial concession operations have not yielded the expected profitability projected during the bidding process. The most extreme example is Madarail, which experienced unfavorable economic and political conditions (i.e. currency devaluation, economic recession and civil disturbances) during the first 18 months of its operations that created much lower revenues and higher costs that translated into a liquidity crunch that made it technically bankrupt. This situation forced a major restructuring of the terms of its concession, with two thirds of the initial USD 40.1 million five-year investment program being financed as a grant by the Malagasy Government.

Even when initial traffic and revenue projections are achieved, other cash flow problems (e.g. unforeseen increases in investments needs, delayed availability of loan funds or non-payment of compensation for passenger service obligations) can quickly trigger a liquidity crisis. Camrail suffered from just such a combination of events during its first five years of operations and it was forced it to borrow short-term funds from local banks, with a sharp deterioration in its debt-to-equity ratio. Transrail, currently is its third year of operations, is suffering from similar problems; outstanding track usage charges owed by parastatal train operators reached the equivalent of 15% of it annual operating revenues by the end of 2005.

Two of the more recent concessions, RVR and TRC, also appear to be in serious financial trouble and some way from achieving their financial projections. with infrastructure investment from the private to the public sector. This reflects the weak financial basis of many of the concessions, whose business fundamentals were (and in many cases still are, in spite of the projections made during the bidding process (Box 7.1)), insufficient to support major investment on a commercial basis and are all too prone to significant liquidity problems. This in turn means major asset maintenance and re-investment is always likely to be a problem. Concession fees are often divided between a fixed and a variable component, generally computed as a percentage of gross revenues. In some cases they are intended to reflect the cost to the Government Authority of providing assets to a concessionaire (as in a leasing agreement) but more generally they are designed to ensure that private operators share the projected financial benefits generated by their businesses with Governments. Successful railway concessions normally also pay a series of taxes (e.g., value added tax, personnel social taxes, income tax) and in many concessions this is greater than the concession fees when taken over the projected lifespan of a concession (Figure 7.9). This diagram is based on the projected concession fees, income tax and net profit margins for eight concessions over their operational lifetime (typically 20 to 25 years). Both concession fees and income tax range from 2% to 14% of gross revenues. Meanwhile, net profit margins for these concessions (defined as total operating revenues minus total operating costs minus depreciation and interest on debt capital minus taxable income) range from 0% (Madarail) to 25% (Zambia) of net revenues. Bid projections need to be carefully interpreted. In the case of Zambia, the ‘fixed fee’ in the concession agreement actually increases steadily through the concession period; however, it is only paid in full if the very ambitious traffic projections included in the reference financial model were achieved to within 3% and would otherwise be adjusted downwards; it is doubtful if the Zambian Government has received anything substantial from this fee. There is also a clear distinction between the well-established Camrail and Sitarail concessions, with their current expectations of modest profit margins and the more optimistic forecasts of the Zambia, Kenya/Uganda and Tanzania concessions. The aggregate of the returns to Government (through concession fees and taxes) and to the concessionaires (in the form of profit) in these three latter cases range from 35% to 50% of net revenue, equivalent to an operating ratio of 50% to 65%. Such a ratio is only achieved by very few railways worldwide, normally much larger with modern plant and equipment and with flows that are much denser than on SSA railways.

65 RAILWAYS IN SSA

Figure 7.9 Concession fees, taxes and projected profits

30%

25%

20% e u n e v e r

t 15% e n

f o

% 10%

5%

0% Zambia Transrail KRC Camrail Beira Sitarail Madarail TRC

Variable concession fee Fixed concession fee Taxes Net profit

Given the relative size of taxes (largely income tax) and concession fees, it is clear governments should consider the combined impact of both revenue streams when negotiating a concession. A concession fee based on net revenue is undoubtedly a more precisely defined measure70 but income tax does provide more flexibility in years in which there are unforeseen difficulties. However, regardless of the mix of fees and taxes, and of any promises made during the bid process, a concessioned railway’s strategy will always be ultimately constrained by the business fundamentals of the proposed railway privatization deal. It is critical that this is clearly recognized by both Governments and advisors during the development of a railway privatization strategy as a concessionaire will only be able to support a finite financial outflow, irrespective of whether it is in the form of concession fees, borrowing costs or rolling stock acquisition costs and proposed concessions with high levels of both debt and concession fees will be prime candidates for renegotiation. Two good examples are Madarail and Camrail: in the case of Madarail, the initial investment plan represented debt by year 5 (i.e. 2008) of 18 cents per net tonne-kilometer, which would have been impossible to service given an average revenue of only 5 c/ntkm. This became clear when in June 2005, after less than 2 years of operations, the Malagasy Government agreed to take over 2/3 of Madarail’s debt in order to reduce its debt to a sustainable level. Likewise, in the case of Camrail, the year 5 debt was about 8 c/ntkm; in 2005 Camrail and the Government agreed on a concession amendment whose primary effect was to transfer, until 2015, the cost of future track financing to the Government and to cap the concession fee to under 4% of net revenue.

70 There are a number of ways in which funds can be extracted by a concessionaire under the guise of costs, such as inflated ‘technical assistance’ fees for providing management,

66 RAILWAYS IN SSA

8 THE WAY AHEAD

8.1 Introduction

Over the last forty years, governments in Africa have abandoned the explicit and implicit subsidies once granted to the rail industry and instead directed most of their transport investment into the roads sector. This has been combined with general liberalization of the economies with long-established parastatal trading organizations being replaced by smaller and more nimble trading groups. At the same time, there was often been little restructuring of the rail sector, other than some attempts to reduce the over-manning that resulted as traffic volumes declined. By the 1990’s, many of the SSA railways, including several that have since been concessioned, were badly run-down, requiring substantial rehabilitation of both infrastructure and rollingstock. They generally carried volumes that are very low by world standards, often no more than a branch-line would carry in many countries. A few railways had substantial mineral traffic but most were carrying semi-bulk freight between the interior and the ports and vice versa; only in a few cases were there significant internal flows. Private involvement in general-purpose railways71 began in earnest in 1992 with the “affermage”72 of the railway operations between Abidjan and Ouagadougou (Côte d’Ivoire/Burkina Faso). Fourteen systems in SSA have now been concessioned or contracted 73 (Table 5.1). Another four are at varying stages of progress. Arrangements in three of the fourteen networks have been cancelled (and subsequently revived with different operators), one has been badly affected by war and one has suffered from natural disasters and long procedural delays. Six have operated for five years or more but only two of these without a significant dislocation of some sort. Except for the railways immediately adjacent to South Africa (Botswana, Swaziland and, to a limited extent, Namibia), those that have not been concessioned have continued to deteriorate over the past decade with the possibility in a number of cases that these declines will be terminal. Many governments in Africa will only consider concessions as a last-ditch solution but in many cases the railways have been left to deteriorate for too long and it will be a struggle to permanently retrieve the situation. The concessions have not been without their problems. It has been difficult in many cases to find more than a very few bidders and their financial resources in several cases have been insufficient to finance the major investments required; as a result the State has had to guarantee investments and, even then, mobilizing finance has been slow. Concessionaires have generally been unenthusiastic about running passenger services, which do not generate the same revenues as freight and tie-up scarce traction-power, but this has not been helped by delays and disputes about the payment of government compensation for non-profitable services. Further problems have arisen about the level of concession fees, the length of the concession and the redundancy payments to the staff no longer required following concessioning. Yet, despite these vicissitudes, the results to date are encouraging, even if not all the prior expectations have been met. Most, but not all, of the concessioned railways have improved both their traffic levels and their productivity and are providing a better service to users, albeit after a solid injection of investment by donors and IFIs. It is, however, arguable that some of this improvement might have occurred in any event given the investment, not all concessions have

71 Excluding railways built and operated by private companies for their exclusive needs (see Box 1). 72 Type of concession contract in which the operator leases assets from the public authority, while the latter provides major investments. 73 This paper uses the term ‘concessioning’ to cover both leasing agreements (affermage) and pure concession contracts, in which the private sector carries both investment costs and commercial risks related to the operation and/or construction or rehabilitation of rolling stock and/or infrastructure for a fixed period. The total quoted includes the SNCC and Togo management contracts.

67 RAILWAYS IN SSA been as successful as, say, Camrail appears to have been, and most have only been in existence for a comparatively short time compared to typical railway asset lives. In addition, responsibility for the on-going rehabilitation and maintenance of track is rapidly emerging as a key issue between concessionaires and Governments. A key government objective in many railway concessions is to obtain finance (whether private or through IFIs) to rehabilitate life- expired track infrastructure. However, for most private operators, track rehabilitation and, especially, track renewal is a major expense (about half the long-term cost of track maintenance) which is both a severe drain on available funds as well as being one which can be deferred (and repeatedly was in the past) at the cost of a few extra speed restrictions and derailments. So are concessions looking like a long-term answer, or are they merely short-term fixes that are living off investment by third parties and which will prove unsustainable in the long-term. And what more needs to be done to ensure a sustainable sector. 8.2 Concession performance

The biggest impact of concessionaires to date has been improving operations; given the weak investment and regulatory climate in many African countries, investment flows have been limited. The nature and size of the privatized transport operations and infrastructure have also required a range of incentives (financial, economic, commercial and regulatory) in order to secure private operators’ interest; the scope and scale of these practices have raised many questions about the actual viability of the completed transactions. Nevertheless, the overall impact of concessioning to date on operations has been broadly positive:  Productive efficiency has clearly improved (Chapter 6). Labor productivity has increased steadily in all the concessions which have operated for over five years and similar figures are likely to come from most of the other recent concessions. Asset productivity has also generally increased.

 Allocative efficiency is difficult to measure directly but the evidence is generally positive. The improved railway productivity, the active searching for new traffic by concessionaires and the improvement in internal business practices have all improved railway cost and pricing structures and, perhaps most importantly, lifted the level of service, thus helping to attract traffic to the mode which can carry it most efficiently and improve intermodal competition74.  In general, with two significant exceptions (Zambia and Transrail), concessionaires have lived up to the passenger service requirements in their concession agreements, even where it has been operationally difficult for them to do so, or where promised Public Service Obligation (PSO) payments have not been forthcoming in practice. However, many of these services are a hangover from previous times and the passengers served would often be far better, and almost certainly more economically, served with a basic road–based system. Concessionaires faced with significant losses on such services are likely to be far more pro- active in pushing for the alternatives to be considered than a government-run rail system.  Few of the concessions are now immune from road competition, except for isolated cases where roads have still to be constructed or there are heavy mineral movements. A 2006 review of four concessions found little evidence of any monopolistic behaviour by concessionaires, either in terms of freight rates or of services being reduced so that resources could be redeployed to favored users, over and above changes in services that any commercialized railway does in response to changing traffic patterns.

74 The combination of technical efficiency and improved level of service has generated particular economic benefits in West Africa; some of the inland countries have bonded warehouses at the ports and transport is then “organized” through a group of accredited transport operators. In the past, these operators have been virtually free to set their own price as the competition from rail has been supply-constrained. The improved technical efficiency automatically expands rail’s capacity, thereby bringing direct pressure to bear on the road operators.

68 RAILWAYS IN SSA

 Similarly there is no evidence that personal travel has been made more expensive for the poor, other than in the case of Transrail services between Bamako and Keyes. While concessionaires generally have a more appropriate cost structure than their predecessors, this is not to say it is the ideal cost structure; operating cost on railways is a function of capital invested as well as operating efficiency and many of these railways have historically been starved of capital where it is required, thus substantially increasing overall operating costs. Whilst rail is generally competitive with road over medium-long distances, this is not necessarily so if rail can only operate at 10-15 km/hr. However, if rail is to compete effectively, this is not just a function of speed – 40 km/hr is often “fit for purpose” – but it also needs to display the other ingredients of good service quality, particularly reliability and security. Probably the single largest disappointment to governments has been the lack of non-IFI-related funding of infrastructure to date. With the exception of the Sitarail ‘affermage’, concession agreements clearly put the responsibility of financing track maintenance and renewal on private operators75. Likewise, all rolling stock financing has also been left to each individual concessionaire under each contract. However, with the exception of the Beitbridge Railway, which relies on “take-or-pay” clauses, Nacala, which was being funded at semi-commercial rates, and Zambia, where the investment program is modest and being funded directly by the concessionaire, most concessions rely heavily upon the on-lending of IFI loans, with below- market borrowing costs, lengthy loan tenors and grace periods to finance infrastructure. Loans have been provided for rollingstock in some cases but for many of the low-volume operators the sensible choice is to find second-hand equipment76. Much of the investment to date has been to address maintenance and renewal backlogs, and without which there would often be no functioning railway. It can thus be characterized as “once-off” investment to get the systems back on their feet. Even the investment that has been made has often been slow to mobilize e.g. over four years in Cameroon and five years on the Nacala line; this is a long time to wait when a business is barely breaking even. There are two major issues: firstly, few, if any, of the concessions have been generating the cash flow needed to make infrastructure investments from their own resources (another shot of investment will be required in many cases, either during the current concessions, or at the end, in order to prepare them for the next concessionaire) and, secondly, there is a clear risk- aversion to investing in infrastructure with a life significantly beyond the length of the concession and for which the concessionaire would receive compensation for the residual value at some time in the future. Even that horizon may be too optimistic; experience in some of the existing concessions shows that even ten years can see the political and economic environment radically changed, leaving the concessionaire high and dry. 8.3 Four key issues

It is becoming clear that classic concession schemes (i.e., those that require the private operator to take on a significant debt burden in relation to revenues) in SSA are unlikely to be financially attractive to bidders other than those which can secure financial benefits not directly linked to the railway operations (e.g. by controlling an the entire distribution chain or through the supply of rail equipment). Consequently, unless the financial structure of SSA rail concessions is changed and/or the market environment in which they operate is favorably altered, the current limited interest shown by private operators in SSA railway concessions will continue. Two key areas are the financing of passenger services and the financing of major track renewals and rehabilitation. These will both need substantial public funding in most concessions. However, if this is provided, governments will also need to strengthen their regulatory capacity to ensure the concession conditions are complied with, as well as ensuring that the impact on the rail sector in

75 Even Sitarail has been financing track renewal as it has had to service the Government debt linked to it. 76 Although this is not always readily available for low-axle-load narrow-gauge lines.

69 RAILWAYS IN SSA general, and concessionaires in particular, is properly considered when policies in other sectors of the economy are being developed. Passenger services As discussed in Chapter 6, few passenger train services are likely to cover even their above-rail costs. Their contribution to infrastructure costs is generally minimal and few services would justify investment in rollingstock, be it loco-hauled or self-propelled. If these are to be operated for more than the initial years of a concession, governments will need to develop a simple compensation scheme, for which the payments need to be made in a timely manner and with a minimum of fuss. Any scheme should be capable of being easily audited and should be reviewed periodically, say every five years; e.g. a scheme in which the concessionaire keeps all the revenue, to encourage him to operate as attractive a service as possible, and is given a contribution per carriage-kilometer (say) towards the cost of running the service. If such schemes are not introduced, passenger services will be a constant bone of contention between the government and the operator, diverting the focus of the concessionaire from the freight services, the improvement of which is of far greater economic importance to the country. Capacity and/or willingness of private operators to finance track renewal: However, it is unclear whether, having been gifted or loaned (at concessional rates) such investment, many of these rail systems will be able to finance major future infrastructure renewals, either through concessionaire injections or from their internally generated returns. The evidence to date is that few, if any, of the concessions are generating significant profits for their operators and certainly not enough to fund long-term renewals. While most concessionaires pay concession fees into general government revenue, none could probably afford to if they were properly accruing funds for future renewals. It therefore remains an open question as to whether a purely privately financed rail concession model is achievable in much of Africa on a sustainable basis. One of the main problems is that track structures have (or should have) lives of several decades, for the traffic volumes typically carried on an SSA railway. If track is renewed with new 43 kg/m welded rail and concrete sleepers, it should have a service life of at least fifty years, and probably more, for a railway carrying up to five million net tonnes (or traffic units) p.a., well beyond the term of most concessions. Even the existing track on some SSA railways has lasted for a hundred years, in spite of being obsolete and without the same strength as current- day material. On a small system, therefore, track renewal is an irregular event which only occurs somewhere on the network every twenty years or so. Conversely, it is almost always possible to defer renewals for several years, albeit at the cost of deteriorating track conditions and reduced operating speeds. For any concessionaire who is uncertain of his long-term future, the safest decision is generally to do as little track renewal as possible. Even if they wish to renew track, and with public-backed debt financing instruments at their disposal, private operators will often struggle to generate sufficient cash-flow to undertake track renewal (costing US$200,000 and upwards per kilometer) in addition to paying concession fees. Few concessions have strong underlying business fundamentals and government objectives to make the level of concession fee and/or rollingstock purchase price the ultimate measure of a successful deal are simultaneously limiting the successful bidder’s ability to renew infrastructure. On a large network, there will be track renewal somewhere almost every year and this can be financed on a regular basis from current earnings; however, when the expenditure may not take place for five or ten years, it is very unlikely that any concessionaire is going to earmark funds on an annual basis and let them just sit in a bank account that far into the future. And raising debt finance for such expenditure (which has very limited resale value for a lending institution) will only be possible as a general corporate loan; this will be almost impossible for a small stand-alone railway which is financially ring-fenced from its shareholders.

70 RAILWAYS IN SSA

This problem is not unique to Africa. When the New Zealand railway was sold in 1993, the infrastructure was in extremely good condition. By 2004, when the infrastructure was bought back by the government, the infrastructure was in poor shape. There were similar experiences in Australia in both Victoria and Tasmania, where the concessionaire also put almost no money into either maintaining or improving the infrastructure before returning it to the state (in the case of Victoria) or requiring the government to provide annual maintenance funding (in the case of Tasmania). Whilst it is desirable that concessions should limit up-front cost to Governments while keeping the financial responsibility of the planned infrastructure investment squarely on private operators, the reality is that few of the concessions actually have the financial capacity to finance this under current policies. Either returns from the concession need to be boosted, or supplementary funding sources developed, or both. SSA railway concessions currently offer two models for financing infrastructure: • Governments finance the initial track rehabilitation and renewal costs, generally by securing specific-purpose loans from IFIs. These loans are on-lent to private operators and tend to only cover the initial five-year investment plan in the hope that they will suffice to propel each concessionaire’s traffic to a level that will enable it to self-finance afterwards the future track investments. This approach is commonly used for railways with a high ratio of initial track investment compared to revenues and which are thus unlikely to mobilize sufficient private financing • Governments do not finance initial track renewal but commit to compensate concessionaires for their investment by the end of the concession (e.g. KRC/URC, TRC and Zambia railways). In such cases the initial amount to be invested is relatively small in relation to expected revenues and private operators are assumed to be able to secure private financing on the merits of their business case. Under both models, Governments usually agree to purchase from the private operators the non- amortized portion of any infrastructure investment they will have financed by the end of their concessions. The ability of many Governments to make such payments is, however, uncertain and this often affects infrastructure investment in the later stages of concessions in other sectors; one solution used by Uganda and Kenya in the KRC/URC concession is to obtain a Partial Risk Guarantee (PRG) from the World Bank to securitize their payment obligations to the Concessionaire. Although this solution considerably strengthens Governments’ track investment reimbursement commitment, the surest way for a Government to secure privately financed track investment remains: 1) to ensure that the Concession (and thus the proposed track investment) are financially sound 2) that the non-amortized value of the assets owed to the Concessionaire by Government at the end of the concession period remains reasonable; and, 3) that the concession agreement allows for a possible extension of the concession period. Many concessions, however, are unlikely to fail at the first hurdle of financial soundness. If the government still wishes to pursue a concession because of the external benefits of rail transport (Chapter 3 showed these can be around one-third of the total benefits of track renewal if a significant volume of road freight transfers following the investment), it will then need to contribute grant funds. One option is to part-finance infrastructure renewal independently from the concessionaire through a Land Transport Renewal Fund, which could be an extension of a Road Fund, funded as a common pool of funds by both the road and rail sectors (for example, concession payments could be paid into the fund rather than into general revenue). A rationale for this could be developed from the external costs avoided by the carriage of passenger and freight by road rather rail (say one-third from public funds).

71 RAILWAYS IN SSA

Effective and efficient regulation of private rail operators Whilst most countries in theory have a counterparty whose responsibility it is to monitor the concession, in practice many concessions ignore many or all of their reporting obligations under the concession agreements. This is caused, in some cases, by operator intransigence (e.g. initially the concessionaire in Zambia flatly refused to acknowledge the counterparty had any authority with respect to the concession), in others by a lack of expertise and/or initiative (in one case, after four years, not one annual report had been filed by the concessionaire nor had this been pursued by the relevant ministry). In such circumstances, it is not surprising that both politicians and bureaucracies are often ill-informed about the problems facing the concessionaire and about the various steps being taken to attempt remedies. Most concession agreements have a long list of requirements with which the concessionaire must comply (there were 34 in the case of Zambia, at least in the initial years of the concession) and letting reporting go by default inevitably creates plenty of scope for argument sooner or later. The main need here is to strengthen regulatory bodies’ capacity as well as imposing annual independent financial and operational audits as part of concession contracts. Funding the regulatory bodies is also a problem for cash-strapped governments. The obvious suggestion is to use the concession fees for this but it is probably preferable not to rely on such an (at times) uncertain source and instead use funding from a Land Transport Fund, if one can be established. As well as analyzing the financial and operational reports provided by railway operators, regulators should also gather information from users (particularly freight customers) to verify tariff and service quality information as well as inter-modal competition. Consistent Government behavior with respect to railway concessionaires, line with good business practices The performance of a number of existing concessions has been negatively affected by uncoordinated actions from various ministries within governments. Examples range from administratively-imposed salary increases to restrictions on access to container facilities and unfunded public service requirements imposed on rail operators. Most of these actions could have been avoided if a properly staffed, funded and authoritative oversight body had existed (the concession counterparty is generally the obvious choice for this). Governments should ensure such bodies have the necessary political and technical powers to control Governments’ actions towards private rail operators. In practice, this means that such bodies should meet frequently (say, monthly) to discuss any pending issues with the concessionaire. They should include, or have ready access to, a railway technical expert, a railway financial expert and a head whose sole work should be to monitor the railway concession, who should report directly to at least the transport and finance ministers. At a more strategic level, Governments should also develop a coherent and realistic policy regarding infrastructure cost recovery. Road operators in Africa, as in may other places, are notorious for overloading with consequent damage to road infrastructure. Road has an articulate and well-organized lobby; counter-arguments from government railways, if they appear at all, have generally been ineffectual and poorly-prepared. The lower road charges are, and the greater the degree of overloading permitted, the lower freight rates by both road and rail will be and the less the funds available from a concessionaire to maintain and upgrade the railway. In spite of these problems, well-run railways still offer the most economical solution to transporting non-time-sensitive general freight on distances over 500-800 kilometers and over much shorter distances for bulk commodities. As such, their revival through concessioning is warranted whenever the business fundamentals supporting this decision are sound. At the same time better solutions must be devised to ensure that while Governments continue to benefit from the substantial economic rates of return from these concessions, private operators’ financial returns are high enough to attract broader and more competitive investor participation.

72 RAILWAYS IN SSA

Annex 1 Rail Networks in Sub-Saharan Africa

Lines (km) Network density Gauge Countries Companies Km/000 Km/ million mm Total Operating km2 pop. Western Africa Benin OCBN 579 438 5.1 66 1000 Burkina Faso SITARAIL 611 639 2.2 39 1000 Ghana GRC 947 947 4.0 40 1067 Guinea Conakry ONCFG 662 383 4.3 104 1000 Bauxite (3) 383 1435/1000 Ivory Coast SITARAIL 611 639 1.9 30 1000.000 Liberia Iron ore (3) 428 0 3.8 124 1435/1067 Mali Transrail 584 584 0.5 51 1000 RCFM 57 57 Mauritania SNIM 704 704 0.7 225 Nigeria NRC 3505 3505 3.8 23 1067 Senegal Transrail 645 645 5.4 77 1000 SNCS/SETINCS 408 408 1000 Sierra Leone MMR 84 0 1.2 13 1067 Togo CFT 492 77 9.2 87 1000 CFTB 30 30 1000 Central Africa Cameroon CAMRAIL 1100 1016 2.3 58 1000 Congo Brazzaville CFCO 795 610 2.3 198 1067 DRC SNCC 3641 2200 2.1 73 1067 CFMK 366 366 1.9 1067 CFU 1028 0 760 Gabon SETRAG 649 649 2.4 428 1435 Eastern Africa Djibouti CDE 100 100 4.3 194 1000 Eritrea 117 0 1.0 21 1000

Ethiopia CDE 681 681 0.6 8 1000 Kenya RVRC 2065 2065 3.8 57 1000 Magadi 171 171 Tanzania TRC 2605 2605 3.8 87 1000 TAZARA 969 969 1067 Uganda URC 1244 261 5.3 38 1000 Southern Africa Angola CFB 1333 246 2.3 222 1067 CFL 479 181 1067 CFMO 907 425 1067 Amboin 122 0 760 Botswana BR 882 882 1.5 443 1067 Malawi CEAR 797 797 6.7 56 1067 Madagascar MADARAIL 723 723 1.5 43 1000 Other 163 163 1000 Mozambique CCFB 987 314 3.9 144 1067 CDN 871 611 1067 CFM 1129 1129 1067 Other 140 0 760

73 RAILWAYS IN SSA

Namibia Transnamib 2382 1683 2.9 1129 1067 South Africa SPOORNET 20247 20247 17.7 440 1067 METRORAIL 1318 1318 1067 Swaziland SR 301 301 17.3 268 1067 Zambia RSZ 1351 1351 3.2 203 1067 TAZARA 893 893 1067 ZR 163 1067 Zimbabwe NRZ 3077 2759 8.3 283 1067 BBR 150 150 1067

74 RAILWAYS IN SSA

Annex 2 – Production Structure (Average 1995-2005)

Transport task Proportion of total task Average haul (km) (million units) Passenger Net tonne- Passenger Freight Passenger Freight -km km Benin OCBN 85 131 39% 61% 160 591 Botswana BR 96 854 10% 90% 182 398 Congo Brazzaville CFCO 178 254 41% 59% 234 436 DRC SNCC 145 407 26% 74% 323 340 DRC CFMK 16 87 16% 84% 98 328 Ethiopia CDE 118 100 54% 46% 229 400 Ghana GRC 124 187 40% 60% 73 152 Kenya KRC 330 1406 19% 81% 633 675 Madagascar FCE 7 1 88% 13% 68 89 Mozambique CFM 58 423 12% 88% 41 155 Mozambique CCFB 40 235 15% 85% 38 281 Namibia Transnamib 49 1000 5% 95% 401 592 Nigeria NRC 131 59 69% 31% 66 699 South Africa Spoornet 1386 104049 1% 99% 326 575 Sudan SRC 104 1250 8% 92% 462 783 Swaziland SR 0 673 0% 100% 0 169 Tanzania TRC 470 1297 27% 73% 647 993 Tazara Tazara 400 822 33% 67% 290 1334 Uganda URC 5 187 3% 97% 155 241

Cameroon Camrail 314 946 25% 75% 245 592 Gabon Setrag 81 1716 5% 95% 392 551 Ivory Coast Sitarail 118 491 19% 81% 360 773 Madagascar Madarail 9 26 26% 74% 145 212 Malawi CEAR 24 61 29% 71% 55 170 Mozamabique CDN 100 143 41% 59% 110 521 Senegal Transrail 269 606 31% 69% 519 1650 Zambia RSZ 186 533 26% 74% 284 337

75 RAILWAYS IN SSA

Annex 3 – Freight Composition for Selected Railway Companies (as % of total tonnage) Cemen Petroleu Ores t and Agric. Fertiliz m and Country Company Timber constr. produc Others Total er product miner materia ts s als l Western Africa Ghana GRC 3 - - - 92 1 4 100 Benin OCBN - 22 - 31 - - 47 100 C.I. – B F. Sitarail - 11 7 23 1 43 15 100 Mali – Senegal Transrail - 9 14 14 2 13 48 100 Central Africa Cameroon Camrail 37 2 4 26 - 19 12 100 Congo CFCO 41 2 1 12 1 2 41 100 DRC CFMK 11 6 - 4 24 - 55 100 Gabon SETRAG 30 - - - 60 - 10 100 DRC SNCC 2 3 - 8 85 - 2 100 Southern Africa Malawi CEAR 23 6 2 5 42 22 100 Ethiopia CDE 12 62 26 100 Zambia RSZ 7 1 13 43 7 29 100 Mozambique All 4 1 52 23 18 100 Tanzania TRC 1 7 - 14 2 27 49 100 Uganda URC - 11 3 5 4 38 39 100 Botswana BR - 14 - 20 37 9 20 100 Madagascar Madarail - 9 - 24 38 14 15 100 Sources: Various

76 RAILWAYS IN SSA

Annex 4: Average Yields (1995-2005)

Passenger Freight Country Company (US cent/pkm) (US cent/ntkm) Benin OCBN 2.0 5.8 Botswana BR 1.3 2.4 Congo Brazzaville CFCO 5.6 10.7 DRC SNCC 3.1 12.5 DRC CFMK 4.2 13.7 Ethiopia CDE 3.1 12.5 Ghana GRC 2.4 4.4 Kenya KRC 0.6 3.8 Mozambique CFM 1.0 3.0 Mozambique CCFB 0.5 3.3 Tanzania TRC 1.6 4.0 Tazara Tazara 1.1 3.0 Uganda URC 2.3 15.2

Cameroon Camrail 2.2 5.2 Gabon Setrag 8.6 2.5 Ivory Coast Sitarail 3.3 5.5 Madagascar Madarail - 5.1 Malawi CEAR 1.0 5.8 Mozamabique CDN 0.9 5.4 Senegal Transrail 2.2 3.3 Zambia RSZ 0.8 3.9

77 RAILWAYS IN SSA

ANNEX 5 RAIL LINKS PROPOSED BY ARU IN 1979 Route Countries 1 a) Bamako – Sissako – Bobodioulasso Mali – Burkina Faso b) Sissako - Ouagadougou Mali - Côte d’Ivoire c) Parakou – Niamey – Ansongo - Gao Benin, Niger and Mali d) Ouagadougou – Dori – Niamey Niger – Burkina Faso 2 a) Ouagadougou – Koupela – Diapaga – Dosso Niger – Burkina Faso b) Anecho - Segboroue Togo –Benin c) Pobe – Ilaro Benin – Nigeria 3 a) Kaura Namoda – Dosso – Niamey Niger – Nigeria b) Zaria – Gaya – Dosso 4 Maiduguri – Ndjamena – Nyala Nigeria- Chad and Sudan 5 Dimbokro - Man - Nzerekore - Tambacounda Côte d’Ivoire, Liberia, Guinea -Nouakchott - Nouadhibou – Marrakech – Senegal – Mauritania and Morocco. 6 Kousoussa – Bamako Guinea – Mali 7 a) Sunyani - Ouagadougou Ghana – Burkina Faso b) Accra – Tema – Lome Ghana – Togo c) Abidjan –Takoradi Côte d’Ivoire – Ghana 8 Blitta-Pama - Diapaga – Dosso Togo, Burkina Faso, Niger 9 a) Yaounde – Bangui Cameroon - CAR b) Dein - Bangui – Belinga Sudan – CAR – Congo – Cameroon – Gabon c) Belinga - Yaounde Gabon – Cameroon 10 Aswan - Wadi Halfa Egypt – Sudan Wau-Suba - Gulu Sudan – Uganda Juba - Mugbere Sudan – DRC Gulu – Arua - Mugbere Uganda – Sudan - DRC 11 Sfax - Tripoli – Salum Tunisia – Libya 12 Salima – Lilongwe - Michindji – Mpika Malawi – Zambia 13 a) Akordat – Tessenei Ethiopia b) Addis Ababa - Nairobi Ethiopia – Nairobi 14 Ilebo -Kinshasa and Brazzaville DRC – Congo 15 Serpa-Pinto (Angola) - Kataba (Zambia) linking Angola Angola – Zambia and Zambia 16 Gobabis (Namibia) and Francis-Town (Botswana) linking Namibia – Botswana Namibia and Botswana 17 Tsumed – Serpa Pinto - Kinshasa - Brazzaville – Namibia, Angola, DRC, Congo Mbinda- Franceville and Gabon. 18 Gao – Abadla Mali - Algeria

78 RAILWAYS IN SSA

ANNEX 6 Reference documents

Briceño-Garmendia C. and Foster V. (2007), More Fiscal Ressources for Infrastructure? Evidence from East Africa, Sustainable Development Department Africa Region, The World Bank, June.

Bullock R. (2005), Results of Railway Privatization in Africa, The Report for the World Bank.

Bureau of Industry Economics (1995), Rail Freight, International Benchmarking, Report 95/22, Australia Governement Publishing Service.

Coelli T. and Lawrence D. (Edited, 2006), Performance Measurement and Regulation of Network utilities, Edward Elgar.

Coelli T., Estache A., Perelman S. and Trujillo L. (2003), A Primer on Efficiency Measurement for Utilities and Transport Regulators, WBI Development Studies.

Estache A., Perelman S. et Mbangala M. (2004), Les techniques de mesure de l’efficience des infrastructures pour les régulateurs d’Afrique francophone, Matériel de lecture.

Estache, A. (2005), How Much Do We Know About Sub-Saharan Africa’s Infrastructure and the Impact of its 1990 Reforms? Mimeo, World Bank.

International Union of Railways (2004), Railway time-series data, World Bank (2002),

Labeau A. et al. (2005), Projet de transport multimodal. Mission d’identification en RDC. Rapport de la Banque Mondiale.

Le Rail (1988), Yearbook of African Railways.

Leeds University (1979), A Comparative Study of European Rail Performance, Published by British Railways Board.

Mazars et Gerard (2005), Appui à la mise en concession de l’OCBN, rapport.

Mbangala M. (2007), Efficience économique des chemins de fer en Afrique Sub-Saharienne in Pichault F. et Nizet J. (Editeurs) (2007), « La gestion des entreprises en Afrique Sub- Saharienne », Editions l’Harmattan.

Mbangala M. (2004), Management of railways in Sub-Saharan Africa. Railway productivity analysis, Rail international, pp. 13-21.

Mbangala M. (2001), L’évaluation de la performance économique et sociale des entreprises publiques africaines des chemins de fer par la méthode des comptes de surplus, Annals of Public and Cooperative Economics, vol. 72, n° 2, June, pp. 183-207.

Mbangala M. (2001), Le transport ferroviaire en Afrique noire. Fonctionnement, Performances, Perspectives, Editions de l’Université de Liège.

79 RAILWAYS IN SSA

Mbangala M. et Perelman S. (1997), L’efficacité technique des chemins de fer en Afrique Sub- Saharienne : une comparaison internationale par la méthode de DEA, Revue d’Economie de Développement, 3/93, France.

Murdoch, Jill. (2005), Assessing the Impact of Privatization in Africa – case study of Camrail

Phipps, L (2008), Review of the Effectiveness of Rail Concessions in the SADC Region, USAID

Pozzo di Borgo P. (2005), Railway Concession Experience in Sub-Saharan Africa, World Bank, Washington, DC.

Pozzo di Borgo P. and al. (2006), Review of Selected Railway Concessions in Sub-Saharan Africa, World Bank Report.

Rail Africain (2005), Magazine d’information de l’Union Africaine des Chemins de fer, 32ème assemblée générale de l’UAC.

Rapports d’activité et Rapports de missions (CAMRAIL, CFMK, NRC, OCBN, SETRAG, SITARAIL, TRC, SOFRERAIL).

Rigaud G. et Allix B. (2006), Consultation multipartite sur la concession de l’exploitation de l’activité ferroviaire sur le chemin de fer Dakar – Bamako, Rapport.

Thompson L. S. (2007), Spoornet and Transnet Sectoral Reference Paper, Thompson, Galenson & Associates.

Thompson, Louis S., Karim-Jacques Budin and Antonio Estache (2001), Private Investment in Railways: Experience from South and North America, Africa and New Zealand, European Transport Conference.

80

Load more

  87

Copy Link