Department For International )FID ~~~Development TITLE: by. J LHine Overseas Centre Transport Research Laboratory Crowthorne Berkshire RG45 6AU United Kingdom PAI 260191 Can time savings be used'?:. Some evidence from the trucking industry in Pakistan PA1260/91 HINE, J L (1 99 1). Can time savings be used?: Some evidence from the trucking industry in Pakistan. International Seminar on Expressways/High Speed Facilities in Developing Countries, Bangalore, India 13-16 June 1991. INDIAN ROADS CONGRESS ANDPERMANENT INTERNATIONAL ASSOCIATION OF ROAD CONGRESSES PARIS 1INTERNATIONAL SEMINAR ON EXPRESSWAYS/HIGH SPEED FACILITIES IN DEVELOPING COUNTRIES BANGALORE, JUNE 13 -16, 1991 CAN TIME SAVINGS BE USED? SOME EVIDENCE FROM THE TRUCKING INDUSTRY IN PAKISTAN BY J.L. HIN'E Reprints from Preliminary Publication VIA 1 CAN TIME SAVINGS BE USED?: SOME EVIDENCE FROM THE TRUCKING INDUSTRY IN PAKISTAN By J.L HINE* SYNOPSIS Truck journey time. savings can represent an important component of the benefits of investment in main roads. However. there has been disagreement amongst various authors. about the extent to which vehicle time savings can be used fully by being translated into extra trips. A vehicle utilisation survey was carried out in Pakistan during 1985-86. Different vehicle activities Were timed and recorded on a continuous basis for period lasting from f ive days to four weeks. It was found that freight vehicles were in active use (travelling, loading, and unloading) for more than 12 hours per day. From the data a series of elasticities were calculated relating trips to travel time and these showed that in Pakistan's conditions. time savings following road investment. are likely to be used fully. I. INTRODUCTION The rapid growth in traffic volumes experienced by many developing countries has, in recent years, led to increasing interest in providing high capacity roads that are primarily designed to reduce congestion and provide journey time savings. Although time savings can represent a major part of the benefits of road CONTENTS investment,the area has attracted comparatively little research in developing countries. This is especially true Page for freight transport which often forms a substantial 1.Introduction 1-4 proportion of long distance traffic. 2. The Valuation of Commercial .. 1-42 Important components of truck operating costs Vehicle Time Savings (particularly the capital costs, labour costs and 3. The Road Freight Transport I. -43 overheads) are time dependent. To appraise correctly Industry in Pakistan high capacity, high speed roads it is necessary to value 4. Survey Procedure .. 14 the savings in costs of these time dependent com- 5. The Pattern of Vehicle Use I. -43 ponents. To do this correctly it is necessary both to measure the amount of time a vehicle spends working 6. The Relationship between Trip I.. 44 (i.e. moving, loading and unloading) and to make pre- Frequency and Trip Time dictions about the extent to which potential time 7. Conclusions ... I-46 savings may be translated. into productive use. Commonsense might suggest that if road invest- ment brings about journey time savings then, on average, trucks should be able to make full use of the time saved by making extra trips. However, there has been disagreement, amongst various authors, about the extent to which this can be achieved. Some studies have even suggested that a large part of time savings cannot be used readily. * Overseas Unit Transport and Road Research Laboratory. Crowthorne. Berkshire. U.K. 1-42 J~~~~~~~~L.L HINE ON Recently, a study of Pakistan's trucking industry was undertaken by the Overseas Unit of the Transport and Road Research Laboratory (TRRL), UK in cooperation with the National Transport Research Centre (NTRC), Pakistan. As part of the study a survey of vehicle time utilisation was carried out. An analysis of the results of this survey are presented in this Paper. Other published reports-and Papers relating to the study are listed in the references. 2. THE VALUATION OF COMMERCIAL VEHICLE TIME SAVINGS In order to value the benefits of journey time savings for commercial vehicles arising from road. investment, it is necessary to make assumptions or pre- dictions about the extent to which the potential time savings may be translated into productive use. Some authors, such as Dawson and Vass (1974). assume con- stant working hours per year in their formulation of vehicle operating costs; this implies that, following a road investment, all time savings will be fully used and that the value of time (for all time dependent costs) will remain unchanged, per unit of working time. With these assumptions time savings benefits may be simply calculated from the predicted time saved and the value of vehicle and crew working time. A common approach is to relate annual time costs to the distance travelled per year. Time savings benefits are then calculated by predicting a change in the annual distance travelled. This may be estimated either by a formnula derived from vehicle speed or estimated *by the user. Neither procedure is very satisfactory. For example both Winfrey (1969) and De Weille (1966) have proposed a formula from which it can be calculated that a one per cent decline in travel time would bring about a one per cent increase in distance travelled. It is assumed that the time spent travelling per day will remain constant. The assumption of cons- tant travelling time is clearly unrealistic for commer- cial vehicles where loading'and unloading time is an important component of the working day. This approach will nearly always tend to overestimate time savings benefits. Other vehicle operating cost model formulations (such as TRR's 'Road Transport Investment Model', the Central Road Research Institute's, 'Road User Cost Study in India', or the World Bank's4Highway Design and Maintenance Standards Model') provide the user with the option to predict utilisation following road investment or to choose one of a variety of models to achieve the same result. The assumption that travel time savings can be readily translated into extra trips has been questioned by Fleischer( 1962) and Thomas (1983). Their empirical investigations suggest that there is an inflexibility in vehicle operations relating to the constraints of drivers' hours and the scheduling of work which will prevent travel time savings from being fully used, particularly in the short-term. It may be thought that the more inflexibility there is in vehicle operations (resulting from the constraints of drivers' hours regulations and acceptable loading and unloading times) then the greater the probability that time savings will not. be fully used following road investment To test this a computer based simulation model was built. Using different input assumptions the results of the model suggested that while additional constraints on vehicle operations could be expected to reduce the average level of efficiency (in terms of reduced working hours) there were no "a priori" grounds to suggest that they would reduce, at'the margin, the probability of using time savings. Comparing different spectrums of journey times the model provided as many cases where total idle time fell (i.e. time savings were more than fully used) as cases where total idle time rose (i.e. time savings were not fully used) following a reduction in Journey times. The exact outcome was shown to be dependent upon the particular trip length distribution and the set of constraints assumed in the model. There are several ways of trying to estimate how time savings may be used following road investment. One approach is to ask operators to guess what their response would be given the predicted reduction in journey times. Another appro ach is to look at changes in utilisa- tion before and after an investment has been made. Without a total survey of all operators using the route it is not known whether other operators are able make use of the time savings. Likewise it is possible that in other situations with only one major route impressive productivity gains could occur with only very small time savings. Thomas carried out an historical 'before and after' analysis as well as cross-sectional analysis to determine how vehicle productivity might change following road investment in Malaysia. For the histori- cal analysis, data was collected on vehicle productivity before and after the new Kuala Lumpur - Karak highway was opened which reduced vehicle trip times by an aVerage of 45 minutes. Although data on the 1-42 CAN TIME SAVINGS BE USED?: SOME EVIDENCE FROM THE TRUCKING INDUSTRY IN PAKISTAN 14 operations of various types of commercial vehicles were collected, no dramatic improvements in vehicle productivity were found following the opening of the new road section. In fact for certain vehicle types the level of trip making per day fell in the Kuala Lumpur area. In the Malaysian cross-sect ional analysis a number of elasticities were derived for different categories of commercial vehicles between trips made per day and average travel time per trip. Most of the data was collected from a roadside interview survey in which drivers were asked what trips had been made in the previous 24 hours. The elasticities found ranged from -0.2 to -0.6, so that if travel time were reduced by one per cent then trip making could have been expected to rise by between 0.2 and 0.6 per cent. This analysis su~ggests that the models which pre- dict a one per cent increase in distances travelled for one per cent decline in travel time (assuming trip distances remain constant) will clearly overestimate the effects of journey time savings. However, to estimate fully the extent to which vehicle working time savings are used it is also necessary to include loading and unloading time within trip times in the calculation of the elasticities. It is against this background that it was decided to develop further the cross-sectional elasticity analysis using vehicle activity data from Pakistan. 3. THE ROAD FREIGHT TRANSPORT INDUSTRY IN PAKISTAN It has been estimated that currently in Pakistan there are in the order of 45,000 trucks in operation, of which about 95 per cent are private owned. General information on private road freight transport was collected from a roadside interview survey of 3500 truck drivers. Three quarters of the trucks surveyed were two- axle Bedford trucks,.14 per cent were two-axle Japanese trucks and the remainder were divided between three- axle rigid vehicles and tractor-trailer combinations. Less than one per cent of the trucks surveyed were owned by a company for its 'own accounC operations. The industry appeared to be entirely organised on a 'hire and reward' basis. Most trucks were 'found to operate on the basis of picking up business where they could and going from job to job as demand required. It was very common for drivers to work away from base for up to three weeks. It was found, that on average drivers of Bedford trucks returned to base after 7 days and returned to their family after 17 days. The driver is responsible for finding the load, for collecting. revenue, and for repairing the truck. When he-returns to base he has to account for the revenue earned and the expenditures incurred. About 80 per cent of drivers are employees and approximately 17 per cent o'~n their vehicle. The remainder own a part share of te yrehicle. Th~re is an extensive network of freight agents who assi st the driver to find a load. In the survey it was found that over60 percent of loads were obtained using agents. In practice vehicles could usually be found for a consignor within one hour. 4. SURVEY PROCEDURE A 'Vehicle Activity Survey' was undertaken whereby different activities were recorded and timed as they occurred on a continuous basis for periods lasting from five days to over four weeks. The data was collected by both survey staff and by cooperative drivers. The survey staff travelled with their allotted trucks con- tinuously throughout the period, if necessary sleeping on board the truck as it travelled. In total over 600 days of useful data were collected, about one fifth of this being recorded by drivers, the rest by survey staff. This comprised 24 periods of data related to conventional two-axle Bedford trucks, seven periods to Bedford tanker trucks and 14 periods to conventional two and three-axle Mercedes trucks. Detailed records of the timing of all movements, rest periods, loadings, unloadings, waiting periods, and repairs were collected. Vehicle stops of less than IS minutes were ignored. Additional data on the distance travelled, costs incurred and revenue earned were also recorded. 5. THE PATITERN OF VEHICLE USE The Vehicle Activity Survey collected data on trucks engaged on a variety of patterns of operation and covered a wide spectrum of trip lengths. The 45 survey periods covered 405 loaded trips and 327 empty trips.. The mean distances recorded for loaded and empty trips were 347 km and 150 km respectively. TO analyse the pattern of vehicle productivity and time use it was necessary, as far as possible, to allocate time spent in different activities into empty and loaded trip periods. For each trip the time spent moving, load- ing, unloading, resting, or under repair was totalled. It was found that on average each loaded trip and each empty trip was composed of approximately 5 and 3 1-43 1-44 ~~~~~~~~~~J.L. HINE ON separate movements (broken by rest periods) respectively. A loaded trip was defined to begin with the start of loading of an empty truck. A loaded trip was def ined to end at the time it was completely unloaded; so a sequence of multiple partial loadings or unloadings were all to be counted as part of one loaded trip. A series of empty movements (made in any direction) were counted as part of one empty trip. No difference was distinguished between resting and waiting in the analysis. TABLE I .MEAN TIMES SPEWrON EACH-- Ac-rlvrrYFOR LOADED AND EmPTY TRIP PERIODS Moving Loading Unloading Resting Repair Total Hrs H-Irs Hrs His Hrs Hrs Loaded Trips 10.8 2.2 1.9 6.5 0.3 21.8 Empty Trips 4.4 - - 11.3 0.9 16.5 Empty Periods BetweenLoaded Trips - - - 7.1 0.6 7.7 Table I gives the distribution of time per trip spent in different activities. In total, trucks were found to be loaded 56 per cent of the time. Rest and waiting time accounted for 30 per cent of loaded trip time, but overall 63 per cent of rest and waiting occurred while the truck ,;as empty. The average duration between the end of loading and arrival at the destination was found to be 14.9 hrs which gives an average loaded journey speed (includ- ing ~Itermediate rest periods) of 23 kph. From the data an estimate of 109,000 km was calculated for annual vehicle travel. Both of these estimates were very close to estimates derived fromi other surveys which were also carried out as part of the study. A high degree of time utilisation was found for the survey vehicles. Most vehicles worked round the clock with activity interrupted only by short rest and waiting periods. Vehicles were found to be moving 40 per cent of the time and loading or unloading a further II1 per cent of the time. Rest periods accounted for 46 percent of the time. Fig. 1 shows how vehicle use changed throughout the day. The most active movement times were found to be between 16.00 hrs and 02.00 hrs. Even at the quietest time of day, 06.00 hrs, 37 per cent of vehicles were work- ing. The most active loading and unloading times were in the middle of the working day between 08.00 hrs and 18.00 hrs. R..Vng U.1..ding looding Moving (E) Mo-ing (L) o 12 14 16 16 20 22 Tin.. (h,$) Fig. 1. Distribution of activities by time of day 6. THE RELATIONSHIP BETWEEN TRIP FREQUENCY AND TRIP TIME In order to estimate the extent to which journey time savings could be translated into extra trip making a series of elasticities were derived from the Vehicle Activity Survey data relating trip frequency to trip time. The survey data was collected over periods lasting up to four weeks. During the survey, some of the truck drivers took their normal 'day offfor recreational rest while other drivers did not. In order to estimate the extent to which this lack of uniformity may have biased the results an additional data set (referred to as the 'adj usted data setf) was prepared which smoothed the time spent on long rest periods. It was also decided to prepare another data set which excluded data from the tanker trucks which have their own unique pattern of operation. The following three data sets were prepared: (i) The basic data set (45 cases) (ii) The adjusted data set (45 cases) (iii) The adjusted data set excluding tankers t38 cases) Each data set was derived from the 45 survey periods of the Vehicle Activity Survey. For each case mean values of number of the trips made per day. and the moving and working times per trip were calculated. It is mentioned above that the Malaysian study calculated elasticities between trips made per day and movement time per trip; the analysis is useful for ind icating how trip frequency (and hence total distance 1 I-44 we 121 X1 CAN TIME SAVINGS BE USED?: SOME EVIDENCE FROM THE TRUCKING INDUSTRY IN PAKISTAN 14 travelled) might change for each vehicle with a change in travel time. An elasticity of -I between trips made and move- ment time is unlikely where loading and unloading times are significant. An elasticity of this magnitude would imply that total working time (including mov- ing loading and unloading time) will increase with travel time savings. The most convenient way of estimating an elas- ticity is by using a log-log regression analysis. The elasticity. is the coefficient Wb in the following regres- sion : Log (trip frequency)- a + b * Log (trip time) Table 2 gives the results of regressions relating trips per day to mean trip movement times for the three data sets mentioned above. Separate regression results are given' for 'loaded trips' and for 'all trips'. 'All trips per day' refers to the mean number of loaded and empty trips made .per day. TAnLE 2. REGRESSIONS RELATING TRIPS PER DAY TO MOVING TIME PER TRIP Basic data set N = 45 (I) Log (all trips per day) =0.682 -0.698 *Log (movement time per (se = 0.036) tlip) R squared va lue = 0.90. T value = -19.6 (2) Log (loaded trips per day) 0.518 -0.709 *Log( loaded movemnent (se = 0.046) time per trip) R squared value = 0.85. T value = -18.1 Adjusted data set. N =45 (3) Log (all trips per day) = 0.715 -0.755 * Log (movement time per (se =0.032) trip) R squared value 0.93. T value -23.7 (4) Log (loaded trips per day) =0.56 -0.768 *Log (loaded movement (se 0.042) timne per trip) R squared value = 0.88 T value = -18.0 Adjusted data set excluding tankers. N = 38 (5) Log (all trips per day) = 0.724 -0.765 *Log (movement time per (se =0.043) trip) R squared value = 0.90. T value =- 17.9 (6) Log (loaded trips per day) = 0.506 -0.702 * Log(loaded movement (se = 0.048) time per trip) R squared value = 0.86. T value =- 14.6 The regressions shown in Table 2 have high R squared values and are very significant The derived elasticities relating trips per day to mean movement time per trip lie within the range -0.70.to -0.77, so that if mean movenleint times are.cut by one per cent then trip making will risq by just over 0.7 of one per cent. These elasticities are much higher than those found in the Malaysian study. To estimate the extent to which time savings can be fully used following road investment it is necessary to calculate the elasticities using total working time including loading and unloading time. It is only in this case that the elasticity coefficients provide an unbiased estimate of the proportion of time savings which will be translated into extra trips. Regressions showing these elasticities are given in Table 3. TABILE 3. RrGRESSIONS RJELATING TRiPs PER DAY TO WORKING TIME PER TRIP Basic. data set N = 45 (7) Log (all trips per day) =_ 0.909- 0.838 * Log (mean total (se =0.047) working time per trip) R squared value = 0.88. T value= -17.7 .(8) Log (loaded trips per day) = 0.873-0.913w Log (mean loaded work (seO= .06 1) time per loaded trip) R squared value = 0.84. tr value = -14.9 (9) Log (loaded trips per day)= 0.90 -0.864 * Log (mean total time (se = 0.045) per loaded trip) R squared value = 0.90. T value = -19.3.. Adjusted Data Set. N =45 (10) Log (all trips per day) = 0.968 -0.913 *Log (mean total working (se =0.041) time per trip) R squared val ue = 0.92. T value = -22.4 (1 1) Log( loaded trips per day) = 0.951 -0.995 * Log(mean loaded work (se =0.054) time per loaded trip) R squared value= 0.89. T value= -18.4 (12) Log (loaded trips per day) = 0.969 -0.932 * Log (mean total work (se =0.038) time perloaded trip) R squared value= 0.93. T value = -24.4 Adjusted data set excluding tankers. N =38 (13) Log (all trips per day) = 1.00 -0.95 'Log (mean total working (se = 0.054) time per trip) R squared value = 0.89. T value = -17.5 (14) Log( loaded trips per day)=-0.88- 0.925 Log( mean loaded work (se =0.066 titne per loaded trip) R squared value =0.85. T value =14.1 (1 5) Log (loaded tri ps pe r day) =1.03 -0.98 * Log (me an total work (se = 0.052) time per loaded trip) R squared value =0.91. T value = 18.7 In Table 3 for each data set three elasticities are presented which relate the following variables: (i) 'all trips per day' to 'mean total working time per trip' (ii) 'loaded trips per day' to 'mean loaded working time per loaded trip' (iii) 'loaded trips per day' to 'mean total working time per loaded trip' The last type of elasticity provides an estimate (as far as is practical in Pakistan conditions) of the res- ponse of round trip' making to total working time per 1-45 1-46 i.L HINE ON CAN TIME SAVINGS BE USED?: SOME EVIDENCE FROM THE TRUCKING INDUSTRY IN PAKISTAN round trip. In the analysis total working time refers to all'empty and loaded moving time and loading and unloading time. Loaded working time refers to loaded moving time and loading and unloading time. No rest, waiting, or repair time is included in the calculation of working time. All of the regressions shown in Table 3 are highly significant and haves high R squared values. The elas- ticities range from -0.84 to -I. Fig. 2 gives a typical plot of the Log-Log relationship (in this case the data used in Regression 12 is used). Overall, by introducing load- ing atnd unloading into the analysis, the elasticity values have risen by 25 per cent. The mean value of the elasticities is -0.92. so that, following road investment, commercial vehicles could be expected to devote over 90 per cent of their resulting time savings to extra trip making. The analysis suggests that the proportions of working time and idle time would hardly change with a reduction in journey times. Hence, in the calculation of commercial vehicle operating costs for Pakistan, it is ,reasonable to assume that time savings would be fully used and that working time per year remaining constant. i 11 A - ' Fig. 2. Loaded trips per day against working time per trip .7. CONCLUSIONS Vehicle time utilisation has been relatively neglected as an area of research and yet it plays an important part in the calculation of benefits arising from journey time savings. To estimate satisfactorily the benefits of time savings following a road 'invest- menit it is necessary to predict the likely change in vehicle utilisation. However, there has been disagree- ment amongst various Author$ on the extent to which time savings can be translated into extra trips. Some studies have suggested that a large part of time savings cannot be used readily. A vehicle utilisation survey in Pakistan found thai freight vehicles were used very intensively. It was found that vehicle running accounted for 40 per cent of the time and that with loading and unloading vehicles were in active use for over 12 hours per day. From the survey data it was possible to derive a series of elasticities relating trip frequency to mean trip working time. The results ranged from -0.84 to -I with a mean value of -0.92. The analysis suggests that follow- ing road investment in Pakistan over 90 per cent of journey time savings of commercial vehicles are likely to be translated into extra trips. ACKNOWLEDGEMENTS The work described in this report forms part of programme of joint research between the Overseas Unit (Head: Mr. J. S. Yerrell) of the Transport and Road Research Laboratory, and the National Transport Research Centre (Head: Mr. M.S. Swati), Pakistan. It is published by permission of the Director, Transport and Road Research Laboratory. REFERENCES I. Central Road Research Institute (1982). --Road User Cost Study in India--. Central Road Research Institute. New Delhi. 2. Chesher. A. and Harrison R. (1987). "vehicle Operating Costs: Evidence from Developing Countries--. Highway Design and Maintenance Standards Series. The World Bank, Baltimore. 3. Dawson. R.F.F. and VassI`. (1974). -vehicle Operating Costs in 1973--" TRRL Report LR 661. Transport and Road Research Laboratory. Crowthorne. 4. De Weille J.. (1966). -Quantification of Road User Savings--. World Bank Staff Occasional Papers No. 2. Baltimore. 5. Fleischer. G.A. (1963). "Effect of Highway Improvement on Travel Time of Commercial Vehicles"-. Highway Research Record. Vol. 12. 6. H-ineJi.L. (1988). -Freight Industry Survey--. National Trans- port Research Centre. Islamabad. Pakistan. NTRC Report 105. 7. Hine,J.L. (1989). -Financing Pakistan's Trucking Industry--. PTRC Summer Annual Meeting. University of Sussex. 8. Hine,.J.L. and Chilver A.S. (1990). *--A Study of Pakistan's Trucking Industry'-. Third International Conference on Sur- vey Methods in Transportation. Washington DC. 9. Parsley. L.L. and Robinson. R. (1982). `The TRRI- Road Invest- ment Model for Developing Countries-, TRRL Report LR 1057. Transport and Road Resea rch Laboratory. Crowthorne. 10. Thomas. S. (1983). --The Value of Time Savings in West Malaysia: Commercial Vehicles"-. TRRL, Report SR 792. Transport and Road Research Laboratory. Crowthorne. 1I. Winfrey. R (1969). "Economic Analysis for Highways". Inter- national Textbook Co,~Scranton. --------- 1 1-46