Are Ghana's roads paying their way? Assessing road use cost and user charges in Ghana

Licy, Research, and External Affairs WORKING PAPERS r Transport l Infrastructure and Urban Development Department The World Bank September 1991 WPS 773 Are Ghana's Roads Paying Their Way? Assessing Road Use Cost and User Charges in Ghana Reuben Gronau The study of road use costs in Ghana showed, first, that such studies are in fact feasible in LDCs, notwithstanding gaps in the data, and second, that they can reveal important inefficiencies in the tax system. The Policy, Research, and Exitemal Affairs Complix distributes PRI Working Papers todissenminate the findmnp of work in progress and to encourage the exchangc of ideas among Bank staff and all others interested tn development issucs. These papers carry the nam of the authors, fonect only their views, and should be used and cited accordinggly. The find.ngs, interprcmauons. and conclusions are the authors'own. They should not be attributed to the World Bank, its Board of Directors, its management, or any of its member countres.  VPlc, Research, and External Affairs| Tirunsport WPS 773 This paper- a product of thc Transport Division, Infrastructure and Urban Dcvelopment Department - - is part of a larger effort in PRE to study and demonstrate mcthods for designing transport user charges and efficicnt transport prices. This research was funded by thb World Bank's Research Support Budget, RPO 674-37, "Transport Taxation and Road User Charges in Sub-Saharan Africa." Copies arc available free from the World Bank, 1818 H Strcet NW, Washington, DC 20433. Please contact Jennifer Francis, room S 10-063, extension 35205 (44 pages). September 1991. Gronau studied how much road damage contrib- deficient in Ghana is the armual licensing fee. utes to road use costs in Ghana and how the Not only shoul(u licensing fees for heavy trucks marginal social costs should be recovered. This be ten times higher than they are now, but required understanding the road deterioration exemptions from the licensing fee should be process better and analyzing the implications for canccled and registration rules strictly enforced. vehicle operating costs and road user charges. Even then, charges on heavy vehicles will The most important thing Gronau learned is not cover costs unless current legal limits of axle that studies of road-user costs are feasible in loading are obeyed. A more cfficient means of reputedly data-poor countries. In Ghana, the reducing the damaging effects of heavy vehicles problem was not so much missing data as lies in structuring the annual fees to reflect how conflicting sources of data. Many of these data much morc damaging two-axle heavy vehicles sources did not exist a few years ago and have are than inultiaxlc vehicles. been established as part of the transport rehabili- tation program. The data sources need consoli- If raising the licensing fee for heavy trucks is dating, but the experience in Ghana proves the not feasible, certainly the govcrnment should feasibility of information gathering and its cancel heavy trucks' exemption from import importance as part of any major transport duties. An import tax of 15 percent and a 10 program. An important component missing in percent purchase tax (the standard rate on Ghana is data on the axle-loading of heavy consumption imports) will go a long way toward vehicles - as different types of axle-load inflict recovering the marginal cost of road usc - and significantly different degrces of damage. will be much harder to evade than the license fee. Gronau found that to bridgc the gap between road-user costs (including the cost of road Gronau found the issue of redistribution of maintenance) and charges, the annual fee for costs and fees of secondary importance in hcavy trucks should be raised tcnfold - to about Ghana, because of the counltry's low fuel con- $800 per vchicle. Fuel taxes alone are not sumption, the current low level of fucl taxes, and adequate to distinguish fully the large difference the fact that expenditures on fuels arc proportion- in road damage costs incurred by hicavy trucks ately the same for the poor and the nonpoor. and private cars. The taxing instrumcnt most iTnc PRE Working Paper Scries disseminates thc findings of work under way in the Bank's Policy, Rcsearch, and External AffairsComplex. Anobjective ofthe series is to get thesc findings out quickly. even if prcsentations arcless than fuily polishcd. The findings, interpretations, and coaclusions in these papers do not necessarily reprcsent official Bank policy. Produced by thc PRE Dissemination Center  Are Ghana's Roads Paying Their Way? Assessing Road Use Cost and User Charges in Ghana* by Re.iben Gronau The Hebrew University of Jerusalem Table of Contents I. Introduction 1 II. Background 8 III. Road User Costs 17 A. Road Damage Externality 17 B. Marginal Congestion Costs 21 C. Road-User Costs 26 IV. The Adequacy of Transport Taxes 31 V. Concluding Comments 36 Appendix 40 References 43 * This research was funded by the World Bank's Research Support Budget, RPO 674-37, "Transport Taxation and Road User Charges in Sub- Saharan Africa."  Acknowledgement This paper is an interim product of an ongoing World Bank research study on Transport Taxation and Road User Charges in Sub-Saharan Africa. The research reported in the paper was carried out with the active involvement of Ghanaian officials and draws extensively on information and data provided by them. The valuable assistance provided by Abbey Sam (Ghana Highway Authority), S. K. Nunoo (Ministry of Road and Highways) and Adote (Ministry of Transport and Comnmunications) is gratefully acknowledged. Esra Bennathan initiated the research with the support and collaboration of Robert Warner and Lyn Squire, and has provided valuable advice and guidance throughout the course of the study. Anil Bhandari contributed with a detailed analysis of the traffic data reported in the paper and provided information on road construction and maintenance costs. William D. 0. Paterson advised on the assessment of road damage and associated cost functions. The report has benefitted from the reviews and comments by David Newbery, Jeffrey Gutman and Thampil Pankaj. The study was managed by Asif Faiz, Highways Adviser in the Infrastructure and Urban Development Department.  I. INTRODUCTION Low population densities and low income per capita put Sub-Sahara Africa (SSA) at a natural disadvantage where investment in transport infrastructure is concerned. Since demand for transport increases with area and with population dispersion, the burden of the investment in and maintenance of the transport network on the economy's resources is inversely related to output per unit of area (GDP per square km, i.e., the product of GDP per capita and population per square km). SSA is in a unique position: low levels of income per capita increase this burden compared with the better-off Latin American countries (e.g., Brazil and Argentina) although the latter are more sparsely populated, and low population densities result in a higher burden compared with densely populated countries in Asia (e.g., India and China) although they are poorer (Table 1). Table 1. Population Density, GDP per Capita and GDP per Square km Population GDP per capita GDP km2 1987 (US$) km2 Sub-Sahara 20.3 340 6,902 Sub-Sahara, excluding Nigeria 16.2 330 5,346 Brazil 16.6 1,960 32,536 Argentina 11.2 2,400 26,880 India 242.6 310 75,206 China 111.7 300 34,627 Japan 322.8 15,800 5,100,240 France 101.7 12,910 1,312,947 Germany, Fed. Rep. 245.8 14,440 3,549,352 U. K. 232.7 10,540 2,452,658 U. S. A. 25.9 18,580 481,222 Sources: Sub-Saharan Africa - From Crisis to Sustainable Growth, Long-term Perspective Study, 1989. The World Bank Atlas, 1989. World Tables, 1991. World Development Report, 1989.  - 2 - As a result, SSA has fewer roads per km2 than Asia oe Latin America, and a larger percentage of these roads is unpaved. Still, it has the highest ratio of roads per dollar of output (Table 2). Crude estimates of the stock of capital embodied in roads, based on a sample of 85 developing countries, indicate that this capital-output ratio (stock per dollar GDP) is substantially higher in Africa than in the other two regions (World Bank, 1988). If one excludes Nigeria, the maintenance of this stock imposes a burden which is twice as high in the Sub-Sahara as in the other two regions. The difference between the regions in the burden imposed by the maintenance of unpaved roads is even higher (by a factor of 5). Table 2. Basic Characteristics of Regional Road Networks Sub- Sub-Sahara East Asia South Latin Sahara excluding & Pacific Asia America Nigeria Number of countries 39 38 7 6 19 Area (mil km2) 20.0 19.1 13.2 5.1 18.9 GNP (1983) (bill. US$) 155.9 85.5 600.4 257.5 583.9 Estimated total road network ('000 km' 1,019 911 1,540 1,680 2,212 Main road network ('000 km) 335 306 451 217 512 Density per 100 km2 1.7 1.6 3.4 4.2 2.7 Density per mil $ 2.1 3.6 0.8 0.8 0.9 Main paved roads ('000 km) 101 79 277 174 252 Density per 100 km2 0.5 0.4 2.1 3.4 1.3 Density per mil $ 0.6 0.9 0.5 0.7 0.4 Replacement value of main network (bill. US$): Paved 25.2 19.9 83.1 31.4 65.5 Unpaved 9.3 9.0 7.0 1.7 10.4 Total 34.5 28.9 90.1 33.1 75.9 Replacement value / GNP: Paved 0.16 0.23 0.14 0.12 0.11 Unpaved 0.06 0.11 0.01 0.01 0.02 Total 0.22 0.34 0.15 0.12 0.13 Source: World Bank (1988), Table A-1.  -3- Road deterioration has been shown (World Bank, 1988) to be a major (currently, perhaps, the most acute) problem of the transport sector in developing countries. Sub- Sahara's situation in this respect is about the same as that of the rest of the developing world. ")n average one quarter of its paved road system is in poor condition and needs major rehabilitation. But the situation in some countries (Ghana, Guinea, Zaire) is even worse: almost two thirds of the system are up for rehabilitation (or abandonment). More important, almost 30 percent of the system (Nigeria excluded) is in only fair condition. This part of the system will soon reach the critical stage where resurfacing is required as periodic maintenance to prevent the system from deteriorating to the point of disrepair and expensive reconstruction. Excluding Nigeria, routine and periodic nmaintenance call for the diversion of about half a percent of domestic resources to prevent the situation from deteriorating. Clearing up the backlog, according to the Bank's estimates, would require close to 5 percent of the GDP. Even if clearing this backlog is spread over 5-10 years, 16 countries (out of the 28 investigated) will -have wv .arshal over 1.25 percent of their resources to get their system back on track. Eleven of the 28 countries will face severe financing problems - even if they increase their road budgets by 50 percent and restrict new construction to one fifth of the new budget, they will not be able to solve their road maintenance problem within the next 10 years. The best way to marshal the additional resources is therefore a problem of utmost urgency in the Sub-Saharan countries. The economic situation facing many of them, however, is not propit,ous for this effort. Twenty-three of the countries experienced a decline in real income per capita between 1980 and 1985: in 11, real income d'_clined in absolute terms. It is thus more important than ever to economize in the consumption of .nfra- structure without interfering with the structural change and the growth of the economies. The power of the price mechanism can be harnessed to this purpose, through an appropriately structured system of road user charges. The recovery of maintenance and capital costs of the inter-urban road system is only one problem that road user charges are supposed to solve. Urban congestion is another. In 1960 only one tenth of the population of SSA lived in urban centers; by 1982 this share doubled. The average growth rate of the urban population in the 1960s was 5.5 percent, and it accelerated to 6.1 percent in the 1970s. Whereas in 1960 only 4 percent of the urban population lived in cities of half a million or more, this share grew to 40 percent in 1980. The number of large cities grew from 3 to 28. The Bank policy study on urban transport (World Bank, 1986, Table A-1) sheds addi- tional light on the urban congestion problem in the SSA. Motorization, though still low, is increasing rapidly. The share of traffic using private cars in Nairobi is about the same as in inost Westem capitals. The speed at which traffic moves in Lagos is only half that of London or Frankfurt. The recent slowdown in income growth has tended to slow down this process, but the rapid expansion of urban population (it is expected to quadruple in the next two decades) cannot but worsen the problems of urban mobility and congestion in Sub- Saharan Africa over the next decade. Road user charges can serve as a major policy tool in addressing this problem.  The major role of road user charges is assigning the right signals to users and operators of the system. One of the salient features of the motor transport production process is the split ownership of the factors entering this process: the government (or some other public authority) builds and maintains the road system, private or public operators own the equipment (i.*., the motor vehicles), and the passengers (or shippers) provide their time. Each of these parties acts according to the prices (i.e., the signals) they face. The passengers base their decisions on the pecuniary cost of travel and their evaluation of their value of time; the firms and households operating the vehicles base their decisions on their costs (and revenues), ignoring the effect their decisions may have on the cost of others; the road authority makes its decisions on the basis of current (and future) traffic volume, regardless of whether this volume is socially optimum or not (often ignoring the travelers' value of time). A prerequisite for an efficient resource allocation is that all parties face the same prices which equal marginal social costs. The two most prominent cases where private marginal costs diverge from the social marginal cost in the road sector (Figure 1) relate to congestion and road damage (other cases such as air pollution, noise, road accidents, etc. are not discussed in this paper). The case of congestion has been discussed at length in the literature (Walters, 1961; Mohring, 1976; and others). The traveler entering a congested route Dases his decision on the cost he faces (including his costs of time) but ignores the cost he imposes on other travelers. Formally, let the marginal private costs of travel (if) equal 7r P + VT(X) where P denotes pecuniary costs, T - travel time, V - the value of time and X - traffic volume. For simplicity it is assumed that only travel time is sensitive to traffic volume (the effect of congestion on operating expenses is dismissed as of secondary importance). The total cost of travel equals 7rX, and the marginal social costs: MSC = 8(irX)/aX = 7r + XV(8T/aX) . The second term, XV(T/aX), is the externality imposed on all other travelers; it equals the change in travel time times the number of travelers. To attain the optimum traffic level the road authority has to impose a charge which equals this externality, A = XV(ff/aX). A second externality, which may be more important in less-developed countries, is the road damage externality (RDE). It has two aspects: the damage caused to the road and the effect the damaged road has on the operating costs (and travel time) of the other vehicles using the road. Engineering data can be used to evaluate the first component of this externality. But it is much more difficult to evaluate, both conceptually and technically, the second component. The impact of road deterioration on the operating costs of the vehicles using it depends on the maintenance strategy. The deterioration of the road and the increased pavement roughness are associated with cyclical changes in the vehicle operating costs (Figure 2). The costs increase as the road deteriorates and decline sharply when the road is repaved and  -5- restored to its original condition. Additional vehicles change the shape of this cycle. When the road authority adopts a maintenance strategy that is condition-responsive, i.e., the road is resurfaced when the road roughness reaches a critical value, R, an additional vehicle moves the whole cycle forwards. The road is going to be resurfaced at time T - AT instead of at time T. Consequently, if an additional vehicle enters the system at time t, all vehicles using the road between t and T - AT are going to encounter increased costs. However, those using the system between T - AT and T are going to face substantially lower costs. The answer to the question whether over the life cycle, operating costs are going to increase or decline depends on the timing of entry, t. The closer t is to the date of the rehabilitation, T, the smaller the increase in operating costs (which may even decline). It was Newbery's (1988) major contribution to show that if the age distribution of the roads (the time elapsed since the last rehabilitation) is uniform (i.e., the distribution of t over the cycle has a uniform distribution), if traffic is the sole source of road deterioration, and if traffic volumes do not grow over time, then on average an additional vehicle imposes no externality on the other vehicles (the positive and negative effects offset each other). Newbery also suggested a formula how to correct this conclusion when weather contributes to the road damage, or when traffic is growing over time. VOC MSC ITXI X tT AT T 2T 3T Figure 1: Private and social cost Figure 2: Vehicle Operating Cost as of traffic congestion Function of Road Deterioration When the timing of overlay is determined arbitrarily (e.g., by budgetary con- siderations), and is insensitive to the state of the road, the externality an additional vehicle imposes on the system is composed of the additional costs of overlay, and the increased operating costs of vehicles using the system between time t and the time of the overlay, T. In this case one cannot escape from estimating the effect of road conditions on vehicle operating costs.  - 6 - Regardless of the maintenance policy, users should be charged for the externality they impose on the system. It should be emphasizedl that from the efficiency perspective there is no difference between the user charge for road damage and for congestion. Whereas the first charge is intended to confront heavy vehicle owners with the right prices when they decide on the type of equipment (e.g., axle configuration), size of shipment (overloading), etc., the second charge should lead to better decisions on whether to travel, and on the choice of travel mode. A second issue is the question how the road authority should finance the investment and upkeep of the road system. It is only natural to question whether an optimum road-user charge system suffices to cover the annual capital and maintenance costs. This question can be shown to be equivalent to the question whether a pricing system based on short-run marginal (social) costs is sufficient to cover average (social) costs. The answer depends on the returns to scale of this 'multiproduct production process'. Given the increasing returns to scale associated with road strengthening, the road user charge associated with road damage will cover only a small fraction of the capital costs of an optimally designed road; if the weather plays an important role in road deterioration, one cannot expect the user charge to cover the total maintenance costs. On the other hand, if one adds the revenues from the congestion charge, the revenue may more than cover capital and maintenance costs, The answer depends on the relative importance of the various components of the road user charge. Even if road user charges do not cover the capital and maintenance costs, it may be decided for motives of 'fairness' or income distribution to finance these costs through transport taxes. The criteria for such taxes, however, should be similar to other taxes. Road-user charges are an instrument for a more efficient rcsource allocation in the road sector and for raising funds for the capital and maintenance costs of the road system. These charges (or their proxies) should be imposed irrespective of the tax policy. Can this instrument be used to solve some of the transport problems of SSA? A prerequisite for implementing this tool is information on the proper level of these charges. This raises a series of questions. What are the essential pieces of information needed for the evaluation of road-user charges? Are the necessary data available in SSA? If certain pieces of information are missing, can they be replaced by information gathered in other countries in the region? To what extent is information collected outside the region relevant to the SSA experience? How sensitive are the results to changes in assumptions? Can one generalize from the experience of a few SSA countries to the region as a whole? The approach used in our research is a case-study approach: we focus on two countries, Ghana and Zimbabwe. The two cases differ in terms of their development status, state of the roads, competitive modes of travel, taxation policy, and enforcement. Hopefully, these differences will also shed light on the experience of other countries. The model for this study is Newbery's work on road-user charges in Tunisia (1988). Our study tries to adapt Newbery's study to the sub-Sahara scene. It aims at simplicity,  providing the decision maker with easy-to-use tools. Simplification is Rften dictated by the nature of the data. The crudeness and unreliability of some of the existing data makes oversophistication inappropriate. The first case study is discussed in this paper and deals with Ghana. Ghana is a typical example of tiie malaise that plagues SSA, and its transport sector is a microcosm of the country's economy as a whole. The extensive road system bulilt before and during the first years of Ghana's independence shared the fate of the rest of the economy', and fell into serious disrepair in the 1970s and 1980s. By the mid-1980s it was found (Kocks Consults, 1986, pp. 3-6) that only 15 percent of the roads could be deemed in good condition, only 40 percent are in fair condition, and over 45 percent are in poor condition - some of them are barely motorable. The condition of feeder roads is even worse: some have 'disappeared' altogether. Glhana has embarked on an ambitious 7-year program of road rehabilitation: to clear the backlog of maintenance and establish a maintenance system that will stabilize road conditions thereafter. At this point it is only natural to ask whether the resources marshaled by government in the form of transport taxes or road user charges will prevent the reoccurrence of this scenario. The questions have been asked before. David Walker conducted a survey of transport taxation in Ghana (TecnEcon, 1988b) with emphasis on the administrative nosts of raising these funds. A more comprehensive study was conducted by Kocks Consultants as part of the Fourth Highway Project (1986). The study tried to estimate tlht size of the road fund required to finance current and periodic maintenance once the maintenance backlog was cleared, and the most effective way of raising these funds. In the absence of data, Kocks did not try to estimate road user costs; instead, they focused on road maintenance costs, traffic forecasts (and forecasts of the vehicle stock and fuel consumption) to derive the tax rates that would generate the target revenue. This study tries to pick up the trail where earlier studies left off. It tries to estimate road user costs, including not only the road damage component but also the cost of congestion (urban and inter-urban). Given these estimates, it evaluates the adequacy of the current transport tax system. The study opens with background discussion of the Ghanaian road sector, followed by the estimation of road user costs. The next section examines the adequacy of the current tax system, and the paper clobs S ith concluding remarks.  - 8 - II. BACKGROUND The Ghana road system consists of 31,700 kms, of which 14,400 kms are trunk roads and 1,800 kms are town roads. Over the years the system suffered from inadequate maintenance, resulting in a deterioration to the point at which, at the time of writing, 36 percent of the trunk roads are defined by the Ghana Highway Authority (GHA) as being in poor condition and 42 percent are regarded as fair; only 40 percent of the rural network are regarded as inotorable. (According to GHA, two thirds of the feeder roads are in poor condition; see Table 3.) Table 3. Distribution of Roads by Region and Condition, 1989 Trunk roads Feeder roads Total Condition (%) Total Condition (%) length lengthl (kin) Good Fair Poor (kln) Good Fair Poor Greater Accra 468 37 45 18 817 1 43 56 Volta 1,511 38 20 42 1,855 11 26 63 Eastern 1,890 23 35 42 2,337 19 36 45 Central 1,409 7 67 26 1,634 12 21 67 Western 1,502 12 48 40 1,076 21 5 74 Ashanti 1,395 24 17 59 3,284 14 5 81 Brong Ahafo 1,839 31 48 21 1,790 16 26 58 Northern 2,790 25 45 30 1,991 15 16 69 Upper East 535 -- 94 6 1,071 8 14 78 Upper West 1,091 7 27 66 1,404 9 11 80 Total 14,430 22 42 37 17,262 13 20 67 Source: Bhandari (1990). The exact number of motor vehicles using this road system is unknown. In the second half of 1987 the Vehicle Examiting and Licensing Department issued roadworthiness certificates to 76,000 vehicles (the number for the first half of 1987 was 68,500, and incomplete data for the first half of 1988 show 67,100 applications for certification). Government vehicles are exempt from the roadworthiness test, and the rules of certification are rarely enforced. The prevailing feeling is, therefore, that the number of certificates issued is nowhere near the real size of the vehicle fleet. Crude estimates prepared by TecnEcon for 1986 (1987, Appendix A-3) claim that only 80 percent of the non-government fleet obtain certificates, and that the government fleet includes over 18,000 vehicles (of which 7,000 are private cars, 5,000 are trucks, and 3,000 are buses). By this estimate, the  total number of vehicles operating in 1987-88 stands at around 110,000. As Table 4b indicates, the vehicle fleet is expanding rapidly: 11,600 vehicles registered in 1985, 12,200 in 1986, and 14,800 in 1987. Customs data indicate that 9,400 cars were imported in 1987 and 12,100 in the first 9 months of 1988. Table 4a. The Licensed Vehicle Fleet by Type and Size, 1987 Type Number Cycles 4,244 Cars 39,952 Taxis 8,396 Buses 0-33 seater 12,794 34+ seater 2,790 Rigid trucks GVW 0-7 t 2,707 8-16 t 2,915 17-20 t 359 22+ t 247 Road tractors GVW 0-24 t 652 25+ t 625 3-axle 261 Total 75,942 Source: Vehicle examination and Licensing Department, July - December 1987.  - 10- Table 4b. New Registration of Motor Vehicles, 1978 to 1987 Tractors Trailers Special Public and Public and pur- ser- mecha- Motor- convey- Goods cara- pose vice nized Cars cycles ance vehicles vans vehicles vehicles equipment Total 1978 8,052 3,011 1,451 3,328 14 783 55 1,024 17,718 1979 8,513 3,528 1,154 2,313 5 366 53 563 16,495 1980 7,283 3,335 1,166 2,079 1 201 81 422 14,568 1981 11,128 1,852 1,764 2,361 69 219 27 827 18,247 1982 5,993 1,616 1,046 1,893 3 237 15 645 11,448 1983 2,908 1,965 1,952 1,864 4 159 26 386 9,264 1984 3,121 2,270 369 1,991 10 83 3 377 8,224 1985 5,192 2,155 852 2,734 40 300 7 319 11,599 1986 4,199 2,793 4,270 7 488 30 385 12.172 1987 6,414 2,589 4,875 16 517 25 413 14,i749 Total 1978-87 62,803 25,114 37,462 169 3,353 323 5,361 134,584 Total 1980-87 46,238 18,575 29,216 150 2,204 214 2,774 100,371 Source: Quarterly Digest of Statistics (1978-85) and Statistical Service Accra (1986-7). The vehicle fleet is very unevenly distributed throughout Ghana. Over half the vehicles obtain certification in Accra; another 7 percent in Tema, and one sixth in Kumasi. Less than one quarter of all vehicles are scattered over the rest of the country. The uneven geographical distribution of vehicle ownership is accompanied by a skewed distribution of road utilization. The Ghana Highway Authority is in charge of preparing the traffic report. The reliability of their data in the past was questionable, and independent counts on certain sections indicate that the GHA figures are understated by 25-30 percent. Steps have recently been taken to ensure greater scrutiny and reliability. Traffic on interurban roads is quite sparse. The 1988 traffic counts (Table 5) indicate that only 14 percent of trunk roads carried over 750 vehicles per day, contributing 60 percent of veh-km (the most traveled 2 percent, with average daily traffic (ADT) in excess of 3,000 vehicles, contributed 15 percent of traffic). On the other hand, the two-thirds least traveled roads (with an ADT of less than 300 vehicles) accounted for less than one fifth of traffic. Most of the heavily traveled roads are surfaced, but only half of them are in good condition. There is a positive association between traffic and the paving of the road and its condition (ADT for paved roads is 720 vs. 150 for unpaved ones; the ADT for good and poor roads are 680 and 290, respectively), but a quarter of the traffic uses unpaved and poor roads.  Table 5. Distribution of Trunk Roads by Traffic Volume and Condition of Road, 1988 Condition of road Road paving Total Good Fair Poor Paved Not paved Average dai ly Length VotUme Length Votume Length Volume Length Votume Length Votume Length Votume traffic (km) ('000 (km) ('000 (km) (l000 (km) ('000 (km) ('000 (knm) ('000 (AAMT) veh-ke/yr) veh-knlyr) veh-kniyr) veh-km/yr) veh-knsyr) veh-km/yr) A. Unknouun 915 0 52 0 610 0 253 0 539 0 376 0 Under 300 8,919 998 1,618 210 3,676 421 3,625 367 1,979 347 6,940 651 301-750 2,665 1,267 584 285 998 475 1,084 507 1,722 844 943 423 751-1500 965 1,052 359 382 443 490 163 180 829 907 135 145 1501-3000 725 1,449 412 848 195 383 118 218 692 1,395 32 55 30(,: -60W- 241 813 112 378 76 251 54 184 241 813 0 0 Total 14,430 5,58o 3,136 2.103 5,999 2.020 5,296 1,475 6,004 4,306 8,426 1,274 S. percent Under 300 66.0 17.9 52.5 10.0 68.2 20.8 71.9 26.1 36.2 8.1 86.2 51.1 301-750 19.7 22.7 18.9 13.6 18.5 23.5 21.5 34.4 31.5 19.6 11.7 33.2 751-1500 7.1 18.9 11.6 18.2 8.2 24.3 3.2 12.2 15.2 21.1 1.7 11.4 1501-3000 5.4 26.0 13.3 40.3 3.6 19.0 2.3 14.8 1Z.7 32.4 0.4 4.3 3001-6000 1.8 14.6 3.6 18.0 1.4 12.4 1.1 12.5 4.4 18.9 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0  - 12 - The skewed distribution of traffic coincides with an uneven geographical distribution (Table 6): whereas the ADT on roads in the Greater Accra region is close to 2,000 ve1icles, in the Eastern, Central and Ashanti regions it is 600-750; the average in the rest of the country is only 200 vehicles (in the Upper West and Northern regions it is less than 100 vehicles). Table 6. Average Daily Traffic by Region, 1980-1988 Region 1980 1981 1982 1983 1984 1985 1986 1987 1988 Greater Accra 2,318 1,103 1,772 n.a. 1,310 1,828 1,823 1,787 1,976 Volta 281 296 269 218 224 257 213 232 297 Eastern 750 831 744 636 492 510 623 656 764 Central 603 668 653 571 594 654 624 628 615 Western 259 229 246 182 190 296 295 327 334 Ashanti 597 594 674 550 647 727 751 739 756 Brong Ahafo 259 230 243 230 250 253 239 221 236 Northern 137 107 94 62 n.a. 75 83 91 82 Upper East. 202 230 171 125 n.a. 129 203 214 182 Upper West. 57 75 101 63 64 n.a. 57 47 82 Totala 417 381 395 288 397 383 382 390 421 Greater Accra 2,318 1,103 1,772 n.a. 1,310 1,828 1,823 1,787 1,976 Ashanti Central 664 714 699 592 567 616 665 676 725 Eastern Other 212 196 191 152 209 197 184 193 208 * Mean of available data. Data on urban traffic is much more limited. A recent study by DeLeuw-Cather International indicates that the average traffic volume on Accra inner streets is about 840 vehicles per hour (slightly higher in the afternoon peak hours and lower at noon). The distribution of traffic in different parts of Accra is far from even. Thus, whereas traffic on major arteries in peak hours reaches 1,600-2,300 vehicles per hour, traffic on non- circumferent city streets is much lower. Accra has by far the most dense concentration of vehicles. Traffic counts in Tema and Sekondi-Takoradi indicate an average hourly traffic volume in peak hours of 450 and 360, respectively.  - 13 - The differences between urban and interurban traffic are not confined only to volume; it is also reflected in the mode distribution (Table 7). Whereas cars and pickups constitute less than 40 percent of interurban traffic, they (including taxis) account for 70 percent of urban travel. The share of buses and trucks, on the other hand, is one half of interurban travel, but only about one fifth of urban travel. The share of cars is even higher in Accra, whereas heavy vehicles play a somewhat more important role in the smaller cities. The volume of traffic (and its distribution) is one of the determinants of the road externality, the other determinant is the traffic loading. The destructive effect of the vehicle on the road paving is an increasing function of the axle load. It is measured in terms of Equivalent Standard Axle Loads (ESALS).' The average ESAL per vehicle depends on the fleet composition (i.e., the share of heavy trucks), the legal axle-load limits and the prevalence of overloading. Early studies (TecnEcon, 1988a) found that overloading on heavily-trafficked routes was rampant and that the average equivalent axle load for 4- and 5-axle articulated trucks exceeded 20 ESAL per vehicle. Later surveys conducted by SWKP (1988a) reported lower values of 11 ESAL per truck. Needless to say, these loadings are much higher than results reported for other countries. Given the high share of heavy trucks in interurban traffic (14 percent), even the low SWKP estimate results in an average ESAL per vehicle of 1.5. More recent studies reported by Bhandari (1990) indicate that the problem of overloading may not be as extreme as originally conceived (Table 8). Still, although the axle load of heavy trucks is comparable to the experience of other countries, their large share results in an average per vehicle which is very high. The 0.9 average, based on Bhandari's figures, is 2.5 times that reported by Newbery for Tunisia (0.38 ESAL). Transport taxes in Ghana are in a state of constant flux. The government employs the whole spectrum of standard taxes - on vehicle acquisition, on inputs (fuel, tires, parts, etc.) - levies license fees, and collects tolls. Although the tax structure changes annually, the overall trend since the early 1980s has been one of lowering transport taxes. The SWKP (1984) estimate for 1981 is that the tax component in the price of private cars exceeded 75 percent, for light commercial vehicles - over 50 percent, for buses - 33 percent, and for trucks - 33 percent to 60 percent. The tax component in the retail price of fuel was 25 percent to 30 percent, and in the price of tires - 40 percent. In comparison, customs data for 1988 reveal that the average effective combined import tax (duties, purchase tax and penalties) per vehicle imported during the first 9 months of 1988 ranged around $450 for pickup trucks and cross-country vehicles and around $720 for saloon cars, i.e., less than 10 percent of their retail prices. The tax component in the price of fuel has dropped to 15 percent (for premium gasoline and diesel oil) and in the price of tires to less than 20 percent. ' The standard unit is an axle load of 8.2t per axle. It is generally accepted that the destructive impact of the axle is proportionate to the fourth power of the axle load [i.e., the ESAL equals (axle load / 8.2)4].  - 14 - Table 7. The Distribution of Traffic by Vehicle Class and Distance Travelled (1987) A. Interurban Percent distribution of VKT Cars Pickups Buses Mammy Trucks wagons Light Medium Heavy Greater Acera 25 16 22 1 1 9 6 11 Volta 20 25 22 14 9 3 7 Eastern 21 16 22 18 8 6 9 Central 20 16 20 15 10 8 1 1 Western 18 15 19 17 10 7 14 Ashanti 22 14 24 13 9 7 1 1 Brong Ahafo 16 19 15 15 13 5 17 Northern 10 15 1 1 14 12 9 29 Upper West 15 28 1 1 10 15 5 16 Upper East 19 27 15 5 13 6 15 National average 19 19 18 13 11 6 14 B. Urban Total Average Percent distribufion of VKT length hourly of traffic roads Cars Taxis Mammy Buses (kmn) wagon and trucks Accra - Inner city 150 800 50 25 10 15 Accra - Outer city 800 440 45 25 12 18 Tema 155 450 40 30 10 20 Sekondi-Takoradi 195 360 30 30 10 30 Note: VKT = Vehicle-km traveled per year. Source: Bhandari (1990).  - 15 - Table 8. Average Axle Loads by Vehicle Type (tons) Vehicle Sample Axle 1 Axle 2 Axle 3 Axle 4 Axle 5 GVW Equiv. type size Std. Axles Pick-up 2 1.6 3.2 4.7 0.03 Light bus 47 2.3 3.6 5.9 0.16 Heavy bus 196 4.0 6.6 10.5 0.79 M-wagon 123 2.7 4.4 7.0 0.21 Light truck 490 3.7 5.2 8.9 0.85 Medium truck 132 4.1 4.8 4.1 13.0 1.13 Heavy truck 192 5.1 7.6 5.6 5.7 24.0 3.98 Heavy truck 34 5.2 5.7 5.4 5.6 5.8 27.6 4.12 Other 2 26.0 7.88 Source: Bhandari, 1990. Current transport taxes have been surveyed in a TecnEcon report for the World Bank (1988b) and will not be replicated here. The main conclusions of the report were as follows: (a) The duty and purchase taxes on imported vehicles at the rate of 25-35 percent (depending on age and engine size) do not exceed - in fact, are less than the rates for other imported consumption goods. (b) Commercial vehicles (trucks, vans, buses) are tax exempt, as are motorized vehicles whose engine capacity does not exceed 1,600cc (if petrol driven) or 1,800cc (if diesel driven). Furthermore, returning expatriates who were away for over two years are exempt, as are lumber companies, foreign construction com- panies and companies with government con.racts. The effective rates are consequently much lower than the official ones, even in the case of high-capacity saloon cars (or cars that are over 5 years old). (c) Vehicle inspections (and licensing fees) are performed once every 6 months (in conjunction with the roadworthiness tests). The fees are based on a progressive scale: US$4-36 for private motorcars, US$13-36 for rigid trucks, and US$71-133 for road tractors. The enforcement of the rules is, however, lax, and government vehicles are exempt. (d) Fuel taxes include a 'road fund' component (about 3 percent of the retail price of gasoline and 5 percent of the price of diesel oil), which is the main source of the fund (other sources aie licensing fees and tolls).  - 16 - (e) Tolls are imposed on the main artery connecting Accra and Tema and on selected bridges. The revenue from this source, however, is significantly lower than one would expect on the basis of traffic polls.  - 17 - m. ROAD USER COSTS A. Road Damage Externality (RDE) As discussed earlier, the analysis of the road damage externality depends heavily on the maintenance strategy adopted by the road authority: whether the timing of the periodic overlay is triggered by the road condition or whether it is fixed. In the first case, Newbery's method allows for an easy calculation of the road damage externality. When traffic is the sole source of road deterioration, when traffic flows are constant, and when the age distribution of roads is uniform, additional traffic has no effect (on average) on the operating costs of other vehicles. In this case, the only externality associated with road damage is the advancement of the resurfacing date. This externality can be measured by the cost of the overlay divided by the number of vehicles that will reduce the road to a state requiring its repair. In this case it is sufficient to know the cost of the periodic overlay and the marginal effect of a vehicle on road condition to measure the externality costs. The effect of an additional standard axle load depends on the standard of the road. The heavier the traffic (both in terms of its volume and loading) the more efficient it is to build a stronger (and more costly) road. The higher the standard of the road, the lower the damage caused by each passing axle. Hence, the volume of traffic and the marginal damage are negatively correlated. Our first estimate of the RDE is based on Newbery's method, allowing for road deterioration due to weather and for growth of traffic. Following Newbery, this calculation ignores the effect of road deterioration on vehicle operating costs.2 Table 9 is based on Newbery's estimates of RDE. The computations are based on a traffic growth rate of 7 percent, a pavement deterioration rate (associated with weather) of 2 percent, and a discount rate of 12 perceni. As the table indicates, road user costs range from 1.76 to 0.36 C per ESAL-km and decline sharply as the pavement loading (and pavement strength) increase. The overlay costs of paved road in Tunisia ($20,000-50,000) are very similar to those prevailing in Ghana, so Newbery's findings for paved roads can be easily adapted by using the Ghanaian loading distribution. In Ghana, the volume of heavy vehicles is lower due to the lower traffic volumes, but in the total loading this factor is more than offset by the heavier axle loads. In Table 9 two estimates of the average axle load are used: the first (based on Bhandari, 1990) of 0.9 ESAL per vehicle, and the second (based on SWKP, 1988a) 2 Newbery has shown that even if one allows for traffic growth the externality associated with increased vehicle operating costs is only of negligible importance (about USCO.02-0.06 per ESAL-km).  - 18 - of 1.5 ESAL per vehicle. In either case, the heavier axle loading in Ghana (compared with Tunisia) contributes to a lower estimate of the average RDE, the estimate being in the range of 0.5-0.6 per EsAL-km (compared with Newbery's estimate of 0.8). Table 9. The Estimation of Road Damage Externality (C/esai-km) Annual RDE Percent of traffic volume ESAL (