PDF content (text-only)
I TRANSPORT RESEARCH LABORATORY TITLE by Some limitations to the opportunities for road investment to promote rural development J L Nine Overseas Centre Transport Research Laboratory Crowthorne Berkshire United Kingdom IA HINE,Jl L (1984). Some limitations to the opportunities for road investment to promote rural development. Intenational Conference on Roads and Development, Paris 22-25 May 1984. PA 1 140/84 .HINE J.L--. Transport and Road Research Laboratory. Grande-Bretagne Limitations des occasions d'investissement routier pour la promotion-des deiveloppemnents ruraux Some limitations to the.,opportun~ities for road investment to promote rural development Algunas limitaciones de las opo'rt~unidades de inversiones viales para promover et desarrollo rural i Une 6tude effectue'e dans la r~gion de 1lAshanti, au Ghana, sugg~re que l'amelioration de la surface des routes aura peu d'effet sur llagriculture.: Il est tres probable que llagricul- ture prof iter~a des invsismnsruir si acons qence de ces derniers est le remplacement du portage par le transport par vehicules. Llapplication d'un mod~le aleatoire du reseau routier, associee a une etude cartographique, suggere que des investissements routiers suppl6mentaires dans la plus grande partie du Ghana, du Kenya et de 1'Inde auront peu d'impact sur le d~veloppement ac~ricole, parce que'les populations rurales de ces pays habitent d~j.~ tout pr~s d'un acces.a des vehicules. A study in the Ashanti Region of Ghana suggests that agriculture will be little affected by the improvement of road surfaces.. Agriculture is most likely 1to benefit from road investment if this leads to the replacement of headloading by vehicle transport. The application 'of a random model of the road network, Coupl~ed with a study of maps, suggests that further road investment in much of Ghana, Kenya and India will have little impact on agricultural develop- merit because their rural populations already live close to vehicle access. Uni estudio en La region Ashanti de Ghana *indica que el majoramiento de superficies viales se harS podo efecto en l~a agricultura. Es el mas probable que la agricultura beneficiary de la inversion vial si ~sta conduce a hacer transportar por vehiculo en vez de cabeza humfana. La aplicacion de un modelo aleatorio de La red vial junto con un estudio de mapas, indica que para gran parte de Los paises. de Ghana, Kenia e India las inversiones adicionales en el sector vial influira poco en e desarrollo agrario, porque sus poblaciones ya se encuentran en proximidades de acceso para vehiculos.. INTRODUCTION This paper is concerned with the circum- stances in which road investment will, in the first i:nstance, *.promot e agricultural devel-opment, and. subs-eque ntly encourage rural 'development in general. The extent to whidh rural populations already have vehicle access is critical.,to any assessment of *the role of road investment to promote rural development. Using a simple mod~el of~the road network an analysis is made of how well the network serves the population and land area. 'The importance of seasonal disruption and the role of toad investment in opening new areas are also discussed.. The Impact of Feeder Roacd Investment in Ghana A study of 33 villages in the:Ashanti' Region in Ghana 1fo'und little'evide'nce to suggest that agriculture was adversel~y affected by inaccessibility~. The villages were between 8 and. 102 km fr~om the 'regional cap~ita'l, Kumasi. All but two of them had vehicle access. Accessibility was measured in terms of the transport costs of moving farm pro- duce to each village's district centre and to Kumasi. If any-thing, the study found that the more remote villages were more agriculturally developed than the more accessible ones. The more remote villages cultivated larger farms, grew more cocoa, sold a greater pro- portion of their produce, and devoted a greater proportion of household labour to farming. This was in spite of the fact that they were at a disadvantage In their ability to obtain loan finance. On the other hand their access to modern inputs such as exten- sionl advice, tractors, fertiliser or cocoa insecticide was unaffected by their inaccess- ibility. Villages with good access gained more of their income from non-agricultural activities,' such as food marketing, food pro- cessing and the provision of rural services. As in other developing countries, agricultural practfice'in dhana is characterised by relatively small volunes of modern Inputs 355 K 14 TABLE 1 Examples of calculated and measured rural accessibilities Ghana Kenya All Kenya India India 'Country (a) (b) C1c) (d) (e) Area (kin 2) 16,000 19,500 571,000 3.3M 3.3M Type of road* R R+T R+T Me R Unit of disaggregation Ashanti 4 39 22 22 Region zones districts provinces provinces Average road density 0.27 0.24 0.11 0.07 0.49 Proportion of population more than 3km from road; (i) predicted by % 20.2 9.48 2 5. 6 57.4 6.7 random model (ii) observed % 0.9 2.52 8.7 60.0 Ratio of predicted Ci) to 22 4 3 1- observed Cii) . Notes: C. a) Bot-h observations -and. Ashanti Region. Cb) Sample: observations identified hut populat random model estimate drawn from maps of and random model estimate drawn from (c) Random model estimate drawn from distri-ct road and track data weighted by district population. Observations based on an analysis of 106 maps ground-weighted by district population and road density data. Cd) Observations based. on 1957 National Sample Survey of the location of villages to metalled roads; random model estimate is from 1961 metalled road length and rural population data. All calculations and estimates based on 2 miles (3.2 kmn). (e) Random model estimate based on .1980 road length data. * R TMe - all roads - motorable tracks ~- metalled roads average density of 0.49 km of roads per km2. This is the highest road density of any of the larger developing countries. It: is only 70 per cent of that of the United States and is higher than that of many European Countries.. The random model suggests that at least 93 per cent of the rural population of India lives within 3km of a road (and at least 98 per cent lives within 5km of a road). The 'all Kenya' estimate shown in the Table suggests. that .8..7 pei~ cent of the rural populatio~n lives further than 3km from a road or motorable track.' However the Major- ity, of th-is 'remote' population (5.5 per cent of the to~t'al) is to, be. found in the- dry north and east of the country where it is possible to. drive over much of the area by truck or four wheel1 drive vehicle. H~ence the absence of road or track does. ,ot necessarily imply a lack of vehicle access. The datai...fr~om- Ghana,. Kenya and India 's'uggces'ts' that in these countries, the rural population lives relatively close to vehicle access. The evidence from other countries is less clear cut. In general, densely populated areas in easy terrain are relatively well served with roads. By contrast, well populated areas in mountainous terrain (as in Nepal or the Andean countries) have relatively poor road access. Medium and sparsely populated areas of desert, savann~a, or the sahel are not well, served with roads although vehicle movement is often not hindered by the natural terrain. The Existing Pttern of. Investment in Minor or Feeder Roads An examination of the road building pro- grammes of a number of developing countries (including Ethiopia - a country with one of the lowest road densitie's in the world) suggests that there is increasing emphasis on rehabilitating and upgrading minor roads and tracks rakther than on providing com- pletely new access. In practice, it seems '358 (eg insecticide, fertiliser) and by a labour force that lives in the vicinity of the farm. Hence agricultural inputs create a relatively small demand for vehicle trans- port. Road investment is only likely to induce a response in production when the costs of moving produce to the market are reduced giving rise to higher farmgate prices. On average transport costs were found to account for between 3 and 6 per cent of the final Kumasi market price for maize, yam and plantain; three of the main staple food crops. Crops bound for Kumasi were transported over distances typically between 120 and 200 km. In order to demonstrate the impact of road investment on the prices received by a farmer at his farmgate, two calculations were made. The first concerned the effect of up- grading a 5 km stretch of motorable track between village and market, assuming that all resultant reductions in vehicle operating costs were passed on to the farmer., It was estimated that the farmgate prices would rise by only 0.1 per cent. The second calculation considered the effect of converting a 5 km stretch of footpath to motorable track between village and market. The consequent reduction in headloading costs was calculated to increase farmgate prices by over 11 per cent - a gain some 100 times larger than the previous change. The Relative Transport Costs of Different Mode s Large changes in transport costs can be expected when human porterage is replaced by motor vehicle. In the Ghanaian example head- loading was calculated to be 12.5 times as expensive as motor vehicle transport. There is much less scope if the change is between animal transport and motor vehicle. Clark and Haswell 2 have shown that transport by animal carts and wagons is typically three times as expensive as by motor vehicle, while in a recent Indonesian study, Rogers 3 pre- sents data suggesting that transport by one- tonne light trucks is twice as expensive as by ox-cart. In practice, the relative costs of different modes depends closely on the size of load and distance to be travelled, there can be substantial economies in vehicle transport if goods are moved some distance and in sufficient quantity. In Ghana, it was estimated that distances were too short and demand too dispersed for motor vehicles to be used between farm and village. It was predicted that savings would arise only if villages formerly inaccessible to motor vehicles were connected to the road n~etwork. The Question of Seasonality in vehicle Access Frequently, rural road investment is justi- fied on the basis of providing "all weather" access. The argument used is that many roads and tracks become impassable during the rainy season and so `all weather' roads are required. Unfortunately the actual extent of traffic disruption is rarely quantified and its effects on the local communities are seldom measured. From the point of view of agricultural production, there are grounds for supposing that the need for substantial road investment to provide "all weather` access is often overstated. In the Ghanaian study, little evidence was found to suggest that~ produce was lost because of impassable roads. In the author's experience it is uncommon for roads and tracks to be impassable to all motor vehicles for more than a day at a time because of seasonal rains. Of course there are exceptions. For example, areas which have many river channels, seasonal swamp land, plastic clay soils, or are subject to con- tinuous heavy 'monsoon' rainfall. In general rainfall tends to both reduce traffic levels and to alter the traffic mix. Saloon cars and buses curtail trips, leaving trucks and four-wheel drive vehicles to cope with the residual demand. Flooding by rivers- in spate. usually diverts all vehicle types for a few hours until the river height decreases. Fares and transport charges often increase in the rainy season; for example, a recent study in Kenya showed that on the "dry weather" roads, fares increased by between 20 and 100 per cent. This was prob- ably a joint response to lower vehicle utilisation, higher variable operating costs and less competition. Nevertheliess, even if transport charges are doubled for the relatively short road lengths 'at risk', in general there will be little effect on the prices paid to the farmer. From the point of view of the agricultural crop cycle, most products are harvested after the major rains and hence in this respect, seasonal disruption may not be that critical. On the other hand, for products like milk, tea and green vegetables that are harvested throughout the year, seasonal disruption may be an important constraint. A Random Model of the Road Network At the local level, transport costs are hig hly dependent on the extent of vehicle accessibility. Hence the quantification of how far rural communities live from vehicle access is critically important when assess- ing the role road investment plays in promoting rural development. This may be achieved by a detailed analysis of maps, although this can be a lengthy procedure when many maps must be analysed., An alternative is to use a model of the road network which will provide a simple, if crude, method of estimating how well the road network serves the land area. The model used here assumes that roads can be represented by a set of infinitely long straight lines, distributed at random on a plane. It can be shown that for an area of 356 'a' km 2with a road length of '22' km, the mean distance (in) to the road network is given by 1X ak m -~ x k The proportion of the area further than a given distance I'di to the network is then given by the formula:- d e mn Thus if an area of 100 sq km has a randomly distributed road network of 25 km in length, then the average distance to the network is:- 1 100 2km The proportion further than 5 km is:- 5 e 2 8.2 per cent. 4 Howe has shown-that-the -number of nodes and links on the main toad network of six devel- oping countries approximates to the number expected from a random road network distri- buted throughout each country. Unfortunately this is insufficient evidence to demonstrate the usefulness of the model for areal cover- age. To validate the model, 22 large-scale maps covering 70 per cent of Ashanti Region were analysed. A set of 25 randomly spaced points was selected for each map and the distance between each point and the nearest: position on the road network was measured. A compari- son between these measurements and the pre- dictions made using the random model is shown in Figure 1 C. .'Id 100, 80 :60 ;40 20 0 2 3 4 5 Distance 'd (kins) .Fig.1' The distribution of area to the rc (Ashanti Region, Ghana) It can be seen that the random model can give a reasonable representation of areal accessibility. However, if the road density in an area is heavily skewed, the area must be split into zones of similar road density and the model applied to each separately. The zones do not have to be contiguous. To test the model in a different region with a wide range of road densities a second study was carried out using 25 large-scale maps of Kenya with average road density ranging from 0..035 to 0.55 km per km2. For the analysis the maps were grouped into 4 zones. Measurement showed that 29.7 per cent of the land area was further than 3 km from the road network, while the random model gave an estimate of 34.3 per cent. If it is assumed that the rural population is even ly spread wi thin each of the zones then an estimate of the distribution of the populat ion from the road network may be calculated by weighting each random model estimate by its respective population. Observed-and random model estimates of the population distribution for the two examples already described are shown in Table 1 together with estimates (based on dis- aggregation by district and province) for the whole of Kenya and India. The Ghanaian and Kenyan estimates shown in the Table suggest that the random model will overestimate the proportion of the popula- tion living more than a given distance from the road network. A much closer relation- ship between predicted and observed is found in the Indian 'metalled' roads example, though there Are strong grounds for supposing that the Indian National Sample Survey over- estimated the distance that the total rural population lived from the network. This is because the observed estimate shown is for villages rather than for population, and it is well known that larger villages are located closer to major roads than smaller villages. In the Ghanaian case, over 90 per cent of the villages with more than 5000 people were located next to major roads while only 13 per cent of villages with 200 people or less were so located. In practice people are often prevented from clustering around the road network because of scarcities of land and water. In the Ashanti Region, with its ample supplies of both, the rural population can afford to be concentrated next to the road netwo rk. By contrast, the large populations of West Kenya and India have to be dispersed over 6 7 8 the available land. In the arid areas of North and East Kenya the population id com- pelled to live close to the water and grazing areas and these are often far from )ad network the road network. Table 1 provides a current estimate. of the distribution of India's rural population in relation to the road network. India now has a road network of 1.6M km giving an 357 .,L tha~t national and internationally sponsored rural road building programmes often follow .the initiatives of local communities and land owners. From- the point of view of rural development, external help should first go -towards pro- viding those facilities' such as bridge crossAings, culverts and other structures that. local' communities cannot provide themselves 'in order to keep basic vehicle access open at a small fraction of the cost of a fully *engineered road. It is only when vehicle traffic increases above a certain minimum that engineered roads are required and these may be sensibly assessed on the basis of *transport cost savings. Opening up New Areas Road investment has a major role to play in ,opening up those remaining areas of the world that are fertile and yet remain relatively uninhabited.: Where land values are low and extensive lmethods of cultivation are practi.ceda, then---- nwah (less adj~acenti to urban :mar;- -kets or- aic~companied by oth er. benefits) 'Will, not appear that attractive. There are many examples of uncultivated fertile land, lying remote from urban centres, yet located adjacent. to road~s. Other development inputs (such as housing, water and land clearing) will .be:,,required if land is to be made more attrac-tivethan existing farmed areas to potential migrants and investors. Conclusions Road building is most likely to encourage agricultural development where human porterage is replaced by motor vehicle transport, but it is unlik ely to have much effect if it me rely improves the quality of road surfaces of existing minor roads and tracks. From the point of view of agriculture the best use of scarce resources is through investment in bridging, minor drainage work and other small scale remedial measures which extend vehicle access and keep routes o pen to motor vehicle traffic. However, major investments designed to improve marginally the 'all-weather capabil- ities of roads may not b e cost-effective for the 'production of most crops, because peak transport demands occur in the drier periods of the year. The application of a random model of the 'road network coupled with map analys~is suggests that the Vast majority of rural populations in countries like India, Kenya and Ghana live close to vehicle access. Hence the impact on agriculture of road building to improve accessibility to the majority of the rural population in these countries is likely to be limited. Road investment has a part to play in the agr~icuiltu'ra~l development of unpopulated but fertile areas. The best results are likely to be achieve~d if road investment is planned in conjunction with other development inputs. ACKNOWLEDGEMENTS This work forms part of a joint research progranmme between the Overseas Unit (Head: J S Yerrell) of the Transport and Road Research Laboratory, U.K. and the Building and Road Research Institute, Kumasi, Ghana. REFERENCES 1.HINE J L, J D N RIVERSON and E A KWAKYE. Accessibility and agricultural development in the Ashanti Region of Ghana. Transport and Road Research Laboratory. Report SR 791, 1983. 2. CLARK C and M R HASWELL. The economics of subsistence agriculture. London, 1964. .3. ROGERS, L H. Tr~aditional1 goods and passenger movements-*in Indonesia. Transportation Research Record No. 89~8. 1983. 4.- HOWE, J D G 1and their use Traffic, Eng. F. Node-link relationships in census evaluation. and Contro~l 15 Oct.* 1973.. (C) Crown Copyright 1984. Published by permission of the Controller of her Britannic Majesty's Stationery Office. 359