Improving the evidence base on journey time reliability on the Trunk Road Network in Scotland
5. Results from the survey of companies
5.1 Survey Conduct
In order to provide a snapshot of the views on reliability of companies and organisations using the Scottish Trunk Road Network, the survey proposed to contact approximately 80 companies by telephone, with the option of online completion for those too busy to reply there and then. Considerable efforts were made to establish telephone interviews, with much "ringing back" but in the end only one telephone interview was successfully conducted. However, several of those approached by phone agreed to complete the questionnaire online.
As part of the construction of a list of names/numbers to contact, a range of organisations that could broadly be described as 'trade bodies' were also approached. Some of these refused to provide individual contacts, but agreed to publicise the on-line survey in their newsletters (and the suchlike) provided to their members. Obviously, there was no control over the timing of those newsletters, and when to expect responses. There were a pleasing number of early responses, but then the rate of responses fell to virtually zero.
At that stage a total of 28 telephone or online responses had been received. As that was only about 40% of the agreed target, a market research company was engaged, which was able to buy in to a large pre-recruited panel. Pleasingly, after some negotiation, agreement was reached to specify a two-way matrix of desired respondents by location and industry sector. Panel members appearing to have managerial responsibilities within their company were favoured. Naturally, not all such grades were available over every combination of location and industrial sector, and there was no control over who would actually respond. In order to be sure of meeting the contractual minima specified, the market research firm chose a pessimistic projected response rate, taking into account that the questions were much more difficult than most that are posed to such panels. In the event, the panel members approach evidently found the questions to be of greater interest than projected, and so 139 responses resulted from this source alone. There were concerns that the respondents might not take the survey seriously, but these proved unfounded.
The total achieved sample size was therefore 167, all obtained between 05/03/14 and 12/05/14. There were two separate questionnaire designs. One was particularly tailored to the effects of unreliability on freight movements and 45 responses to this Freight (F) questionnaire were achieved. Of these, only 33 were judged to be sufficiently complete to be analysed. Only four of the freight respondents was actually a road haulier, i.e. engaged in third party 'hire & reward' work. Others answering the freight questionnaire were mainly those companies involved in 'shipping' (i.e. sending out) freight, either on 'own account' or via a road haulier. Some respondents were receivers of freight. Those respondents who had no significant freight movements to report completed the second, Non-Freight (or NF), questionnaire. This looked at impacts of unreliability on the organisations' staff and customers. It attracted 122 responses, of which 116 were judged to be sufficiently complete to take forward. The total useable sample, to both questionnaires, was therefore 149.
5.2 Results from the Freight (F) Survey.
The spread of Freight respondents over company 'sector', self-chosen as best describing their company's activities, is shown in Table 5.2.1. Almost a third (10) of the Freight respondents are Manufacturers. With another 5 involved in Energy or Construction, about half of the respondents are clearly involved in production. A further 3 companies report themselves as Food and Drink (but not Retail or Services) so they may also be producers. Four respondents are broadly involved with transport and distribution. Services, Retail, Financial and Creative/Digital account for 8 companies, leaving 3 Government respondents.
Table 5.2.2 shows the distribution of the freight sample by location. Two respondents claimed to have company premises all over Scotland, and one claimed to be in Northampton. Over half of the sample is in the 'central belt', but there is good coverage of the north, though not many in the south west of Scotland.
|SECTOR BEST DESCRIBING COMPANIES ACTIVITIES||NUMBER OF RESPONSES|
|Warehouse and Distribution||1|
|Food & Drink||3|
|NEAREST CITY/TOWN||NUMBER OF RESPONDENTS|
Due to the small sample size, splitting respondents into groups for analysis will not generally be robust. Cross-tabulating firms by sector and location, for example, might also permit the identification of some firms; which respondents had been led to believe would be protected against. Anonymised data has been prepared for passing to Transport Scotland.
Table 5.2.3 shows a large range of size of operation, measured as the total tonnage for all that firm's flows p.a. Responses were obtained from 23 firms. A couple hardly moved any freight at all. Typically, several thousand tonnes were moved p.a., and that is reflected in the reported Median. The distribution of tonnages was heavily skewed, so that there is a long tail of high tonnages. Consequently, the Mean tonnage is three times the Median, being around a quarter of a million tonnes p.a., probably involving around 50 lorry loads per working day. The biggest annual tonnage, from a seaport, was almost three million tonnes p.a., no doubt involving over 300 lorry loads per working day.
In order to introduce the concept of Journey Time Reliability, respondents were asked to indicate their feeling regarding whether travel times on the Scottish Trunk Road Network had become less or more reliable over the last 5 years. Table 5.2.4, shows that the majority indicate that the reliability of travel times has become worse. However, it is more than likely that those feeling that way will be over-represented in a survey of reliability with such a large element of self-selection in as in the present case. Stratifying the sample would not have helped, since there would be a divergence of opinion within any feasible strata, and those concerned about reliability within each stratum would be more likely to respond. Hence, it would be unwise to place too great a weight on this result.
|PERCEIVED CHANGE||NUMBER OF RESPONDENTS||%|
Respondents were asked to provide data on a typical freight flow. Table 5.2.5 reports their responses when asked about the nature of the freight concerned. Note that, while respondents generally gave just a single answer, the listed categories are not mutually exclusive. For example, pallets can be containerised. However, it does appear that the sample includes a range of natures of freight. It is not thought surprising that Palletised accounts for a third of the total, as that is the standard form of moving loads of individually small items, e.g. retail goods.
|NATURE OF THE FREIGHT||NUMBER OF RESPONDENTS||%|
|Carried in Tankers||2||6%|
|Metal loads etc.||3||9%|
Table 5.2.6 summarises the 26 responses received to the question of tonnage shipped and lorries dispatched per year for the typical flow. Two respondents did not give tonnage, and another couple reported such a small flow that excluding them was considered, but they are in. Some responses were vague, so some judgement has been used, and somewhat rounded figures given. The load per lorry cannot be determined from the table, but varied between 0.125 tonnes and 30 tonnes. Looking at Table 5.2.6 shows a wide spread of tonnages and numbers of lorry loads shipped in a year. Very occasionally, the traffic moved as a part load. The typical (median) flow was of roughly a lorry load per working day, and carried around 5000 tonnes. Both distributions were highly skew, so that the means were more than 6 times higher than the medians. Seven flows reported more than 1000 lorry loads p.a. and three flows were over 100,000 tonnes p.a., the highest being 500,000 tonnes p.a.
|Lorry Loads p.a.||5||275||1700||25000|
As the survey specifically refers to trunk roads, the typical journeys reported are quite long, taking a mean of 7.3 hours and a median of 4 hours. Table 5.2.7 shows the distribution of Scheduled Journey Times, the shortest being 30 minutes and the longest being 48 hours. No response was obtained from 5 firms.
|Less than 2 hours||4||14%|
|From 2 to less than 4 hours||7||25%|
|From 4 to less than 6 hours||8||29%|
|From 6 to 10 hours||4||14%|
|From 10 to 20 hours||1||4%|
|From 20 to 30 hours||3||11%|
|From 30 to 50 hours||1||4%|
Table 5.2.8 shows the distribution of arrival times, relative to the scheduled arrival time, for the typical freight flows. Each respondent was asked to give a distribution for their flow (egg. 50% arrives On Time, and the remaining 50% arrives within 30 minutes), and Table 5.2.8 shows those figures averaged over the 27 respondents who answered this question. For some receivers, on time deliveries are a must (egg. supermarket distribution centres), and so some slack time is built into schedules. It should not, therefore, be surprising to see 55% of arrivals on time, but it is certainly impressive that another 35% arrive within the hour (given the average journey time was seen in Table 5.2.7 to be 7.3 hours). Of the remaining 10%, half arrives within a further hour, giving 95% within 2 hours. The remaining 5% is spread down quite a long tail. From that distribution it is possible to calculate the mean lateness as 40 minutes and the standard deviation of lateness as 3 hours. That latter figure looks high, but results from the very long tail of the distribution. Median lateness is close to 30 minutes.
|ARRIVAL TIME RANGE||MEAN PERCENTAGE|
|Up to 30 mins. Late||24.5%|
|Between 30 & 60 mins. late||10.5%|
|Between 1 & 2 hrs. Late||5.1%|
|Between 2 & 4 hrs. Late||2.4%|
|Between 4 & 12 hrs. Late||0.8%|
|Between 12 & 24 hrs. late||1.0%|
|Over 24 hours late||0.8%|
Table 5.2.9 reports the stated relationship between the respondent's company and the typical flow of freight. Four respondents did not provide a response to this question, and the quoted percentages ignore those firms. The company is the shipper in about 80% of the cases where an answer was provided. In 34% of cases, the company only shipped the goods, with someone else proving transport. In 38% of cases, the company both shipped and transported the goods. In another 2 cases (7%), the company was both shipper and receiver, i.e. the load moved between company premises. In 14% of cases the company was only the Carrier for the freight. Those 4 cases include the Road Haulier shown in Table 5.2.1, plus a Warehouse/Distribution firm, and two firms in the Services sector. The final two responses were the Seaport and a case where the respondents company was the Receiver only.
|RELATIONSHIP TO FREIGHT||NUMBER OF RESPONDENTS||% OF REPLIES|
|Shipper & Own Account Carrier||11||38%|
|Shipper & Receiver||2||7%|
Respondents were asked to estimate how many management hours were taken up dealing with the consequences of a single lorry load of this traffic arriving a day late. Ten respondents had "No idea". The remaining 23 respondents reported a total of 35 hours, so an average of one and a half hours. Naturally, delays as large as one day would be rare, but it can be deduced from the data that a 4 hour delay would not require more than 90 minutes management time, on average, and that 15 minutes might be a reasonable value to take. STAG 2012 (see Transport Scotland, 2012) reports work values of time at around £40/hour (for car travellers) in 2010, reflecting the gross wages of such travellers. Making the appropriate adjustments to 2014 prices and income levels raises that to about £50/hour. Assuming that the management hours referred to in the survey were fairly senior, it was thought sensible to use here a gross wage rate of £80/hour. Hence 15 minutes of such time would imply a cost of £20 from a 4 hour delay. In the absence of better estimates, this provides a very rough estimate of one component of the cost of unreliability (£5/hour of delay).
Another question asked about the effect on Buffer Stocks of a sustained reduction in the incidence of late arrivals. Five respondents did not provide a reply, and 22 more said there would be no effect. The remaining 6 respondents reported an average of 16% reduction in Buffer Stocks, with all responses being in the range 10% to 20%. Some of the non-respondents would have found the question irrelevant, for example the seaport and those merely involved in carrying the goods. At least one of the firms reporting no reduction had no Buffer Stock to begin with. If were assumed that about 18 firms would have no reduction, then averaging these with the 6 reporting a 16% reduction would give an overall 4% reduction in Buffer Stocks. The value of that will obviously vary from case to case, not least with the value of the goods concerned, but 4% could be a quite significant saving in the working capital involved.
Table 5.2.10 reports the greatest impact on reliability of freight movements is felt to be that roads are insufficiently large to cope. This is closely followed by the concentration of traffic at particular times. Road works came in third place, with Road Design not far behind that. Weather and Accidents were not felt to be particularly important in themselves. The closeness of the Mean and Median results suggests that these views are shared over different sections of the sample.
|IMPACT RATING (out of 10)||MEAN||MEDIAN|
|Roads not large enough||7.7||8|
|Traffic concentrated at particular times||7.3||7|
Table 5.2.11 shows the cost to the firm, per shipment of the typical flow, if road conditions were such that the scheduled journey times had to be increased by stated amounts. A lot of firms failed to identify any cost, possibly because they used Road Hauliers who charged by distance, not appreciating that such charges would be bound to rise, to cover driver wages and vehicle provision for more hours. Most respondents did report some costs for journey time increases of 60 minutes or more. The Median cost given for 60 minutes was £15, rising to about £80 for 2 hours, and rising linearly above that. For 2 hours increased journey time and more, these amounts are above drivers' wages, and several reasons for the costs were given. In some cases penalties were set out in a contract, though if journey times were scheduled to be longer a different contract would surely be drawn up. In other cases additional overtime was mentioned (egg. for warehouse staff), and in one case vessel docking charges were mentioned. The latter were very high compared to most other costs considered, and will have contributed to the Mean costs being much greater than the Median costs.
|30 MINS LONGER||60 MINS LONGER||2 HOURS LONGER||4 HOURS LONGER|
The mean costs start out at near £70 for the first 30 minutes, then increase their rate per minute, so that 60 minutes extra journey time is costed at £170, and 120 minutes at over £500, but the cost per minute then falls, giving £850 for 4 hours. The view was taken that the cost of delays, rather than a scheduled journey time increases, are being picked up. It may have been that respondents interpreted the word "scheduled" as implying that the trip would be rescheduled for that amount of time later, therefore resulting in late arrivals. Accordingly, the responses to this question were amended, inserting a minimum £16 per hour for firms engaged in moving traffic (in respect of driver's wages) and deleting all costs that would only apply to unscheduled time changes. There were so many of the former that they constituted the median, while the mean figures were much reduced. This is shown in Table 5.2.12. The figure of £16/hour is consistent with the figures in Table 9.15 of STAG 2012 (see Transport Scotland, 2012) for "Values of Time per Vehicle". Those figures are £13/hour for HGVs (there called OGVs) and £15.60 for LGVs, in 2010 values and prices, at market prices. Adjusting for forecast income growth (3.3%) and actual inflation (14.4%) over the 4 years to the time of the survey (as just one example of what might be done) raises those figures to £15.35/hour and £18.40/hour for HGVs and LGVs respectively. There is no one exact answer, but £16/hour is clearly a reasonable rough estimate.
|30 MINS LONGER||60 MINS LONGER||2 HOURS LONGER||4 HOURS LONGER|
Table 5.2.13 looks not at scheduled time increases but at unscheduled ones, i.e. unexpected lateness. Two main influences are at play. Firstly, as the delay is unexpected, some contingency expenditures may need to be made on all days (even when there happen to be no delays) in order to manage efficiently on days when such delays do occur. Secondly, since delays, as was seen in Table 5.2.8, do not occur on all days, the delay cost averaged over all shipments will be much less than for a single late shipment. That said, the results in Table 5.2.13 are something of a puzzle. Even more firms can find no costs arising from occasional lateness, presumably thinking that their Road Hauliers have built in some allowance for delay costs and will not be raising their charges in the light of increased delay. In any event, the Median figures are very small. The Mean costs follow those for extra scheduled delay (Table 5.2.11) closely, suggesting that respondents would deal with both forms of delay similarly. It should be noted that the sample mainly contained shippers, rather than receivers, so costs may have fallen elsewhere.
|30 MINS LATE||60 MINS LATE||2 HOURS LATE||4 HOURS LATE|
As discussed in relation to Tables 5.2.11 and 5.2.12, it was again felt appropriate to amend the data to impose a minimum £16 per hour cost for those (many) firms involved in carrying freight, as was done in deriving Table 5.2.12. The equivalent table for Unscheduled Journey Time increases is Table 5.2.14. The differences between Tables 5.2.13 and 5.2.14 are not large, so the amendment has had minimal impact in this case. The amendment was made principally to be consistent with Table 5.2.12 and the analysis provided for the case of Scheduled Journey Time changes.
|30 MINS LATE||60 MINS LATE||2 HOURS LATE||4 HOURS LATE|
From Table 5.2.14, the median figure of £23/hr of lateness will be used as a starting point for the calculation of the reliability ratio, RR. The STAG definition of RR for Road was given (in rewritten form) in section 4.1 as:
RR = (Value of ΔT) / (Value of ΔT) = VOR / VOT
It was decided to work with medians as there were worries about some of the higher costs reported (egg. the vessel docking charges) which have raised the mean greatly above the median (which will have been little affected by outliers or mistakes in the data). The median Values of Time (VOT), for a one hour longer journey, are reported in Tables 5.2.11 and 5.2.12 as £15 and £16 respectively. The choice of which will make little difference, the latter being chosen. Table 5.2.8 gives the reported distribution of lateness, and it was noted that the standard deviation of the reported distribution of lateness was 3 hours (181.6 minutes). In order to estimate VOR, two extremes were considered. The first added a fixed amount of additional lateness to all arrival times in the distribution. This, obviously, left the T unchanged. Secondly, all lateness amounts were increased in proportion. This had the disadvantage of leaving all on-time arrivals as still on-time. Following experimentation, it was decided to take a weighted average of these two extremes: roughly 25% fixed and 75% variable. Experiments included a 20 minute increase to lateness, which raised T by 60 minutes, and a 60 minute increase to lateness, which raised T by 180 minutes. In both cases the resulting distributions of lateness looked plausible. Having checked for a 'scale' effect, it was deduced that one minute of lateness increased SD of lateness by 3 minutes. Working with values per hour, it was concluded that an extra hour of SD of lateness was valued at the value of 20 minutes of lateness. Therefore, one third of the (median) value of 1 hour lateness (£23 in Table 5.2.14) is the value a standard deviation of lateness. Hence, RR can be calculated as follows:
RR = (23/3)/16 = 0.48
There are a number of important caveats that need to be stated. Firstly, only 18 firms provided data on their value of lateness, and that is too small a number to be more than indicative. Secondly, the Value of Time figure is probably somewhat too low, resulting from the decision to work with medians. It is still felt that the original mean value of 60 minutes reported in Table 5.2.11 as £170 is far too high, the revised figure (£21) reported in Table 5.2.12 is equally well based as the figure used (£16). Replacing £16 by £21 would reduce RR to 0.37, not that big a change.
In order to try to throw further light on this matter, and at the risk of working with even smaller samples, it was decided to investigate 'within observation' calculations (i.e. time distributions and valuations by respondent), from which means and medians could be taken, either for the whole sample or for subgroups. Due to sample size considerations, only two subgroups could be identified: Transporters (i.e. Own Account operators plus hauliers) and Others (ie. everyone else, mostly shippers not undertaking Own Account operations). It should be noted that results reported below are for all respondents in those groups answering the stated questions, rather than just those that answered all of the relevant questions for the calculations undertaken.
Table 5.2.15 presents a mixed picture regarding arrival times. Using the median figures (of the Mean and Standard Deviation of the individual arrival time distributions by respondent) gives typical movements arriving 14.5 minutes late with a standard deviation of 18.4 minutes (indicating a long tail). The two groups of respondents do not differ much, in that regard. Taking mean values (of mean and s.d.), however, shows average lateness at 42.1 minutes (consistent with Table 5.2.8), with a standard deviation of 75.9 minutes. Both figures are pulled up by the 'Transporters'. None of the standard deviation figures comes remotely close to the 3 hours, from Table 5.2.8, used above when calculating RR. The 3 hour figure arose because data from 4 firms reporting lots of 24 hour late arrivals was averaged to say all firms had some arrivals 24 hours late. Working with the distributions of individual firm data gives just 4 very high figures among 27. The remaining 23 dominate, so the SD is estimated at roughly 60 mins and 90 mins for the 2 groups, and at 76 minutes rather than 180 minutes for 'All Respondents'.
Table 5.2.16 considers valuations for Scheduled and Unscheduled journey time increases, but only for 60 minutes, with valuations for other amounts of delay ignored. Starting with the median scheduled delay (VOT), it can be seen that the £16 assumed figures for drivers' wages dominates for Transporters, but 'zero' dominates for those who do not do transporting. The mean values, however, go the other way, indicating some high disutility of longer journey times by 'Others'.
Moving on to Unscheduled journey time increases (VOL), Transporters have much lower values than the Others. As was found for 'All Respondents' (repeated from Table 5.2.14), the Means are much higher than the medians. Following the reasoning given earlier for the calculation of the overall RR from the distribution of arrival times pooled over all respondents, VoR figures were obtained by dividing the VOL figures by 3.
Table 5.2.17 uses figures from Table 5.2.16 to derive RR values. As previously discussed, working with means gives very high values for RR, which cannot be said to derive directly from responses to the survey since implausibly high mean VOT figures were replaced. Had that not been done, the reported high values would have been avoided. There can therefore be no complaint about dropping the 'mean' estimates.
Turning to the 'median' estimates, there is the problem that the zero estimate for 'Others' VOT leads inevitably to an estimate of infinity for that RR. It was felt that the best course of action was to apply the relativity of 'Others' to 'All' RR values using means to the 'All' estimate using medians, i.e. 0.48*(4.01/2.7) = 0.71. The preferred RR estimates for Freight are therefore:
FIRMS CARRYING GOODS RR = 0.21
FIRMS JUST SHIPPING OR RECEIVING GOODS RR = 0.71
ALL FIRMS INVOLVED WITH FREIGHT MOVEMENTS RR = 0.48
These values fit well with the literature, see Table 4.4.1, where Carriers have been found to have very low values of RR, while Shippers have much higher values. Note that the overall "All Firms" result merely reflects the mix between the two sub-groups in the sample, and so has no particular significance. Presumably by pure chance, it is equal to the result for the whole sample when pooling over all respondents. Naturally, the previously stated caveats still apply. If more robust estimates are required, then a much larger sample would be necessary.
5.3 Results from the Non-Freight (NF) Survey
Table 5.3.1 shows how the 116 respondents to the Non-Freight (NF) Survey described their position within the organisation. Their descriptions were free-form, so there has been some grouping and simplification to arrive at the 16 categories listed. The most popular of those categories (with 26) was the one covering a myriad of types of "manager", including some "senior" ones, but excluding General Managers (another 3), Financial (another 4) and Managing Directors (another 6). A surprisingly large number, 21, of respondents claimed to be Owners, Partners, or Proprietors. A further 16 described themselves as "Directors", 2 as CEOs and 4 more as Chairpersons. The above accounts for 82 of the 116 respondents, without yet getting to the more technical and specialist grades. It is clear that the sort of respondents targeted have been reached.
|POSITION IN COMPANY||NUMBER OF RESPONDENTS|
Table 5.3.2 shows the reported company location for the respondents. There is a good coverage of all parts of Scotland. There is certainly no shortage of respondents in the north of the country. Compared to the Freight Survey, the Edinburgh area response has overtaken that from the Glasgow area, with both being substantial. Where responses lie between those places named in a given category they are not necessarily reported in the table. The descriptions should therefore be taken as areas, rather than a full list of places mentioned. Surprisingly, there were 7 responses from England and one each from continental Europe and the USA. These have been left these in the analyses reported below.
|NEAREST CITY/TOWN||NUMBER OF RESPONDENTS||CODE|
|South West: Ayr/Irvine/Dumfries/Stewarton||7||AY|
|Other: Nationwide/Outside Scotland/No Answer||13||OT|
Table 5.3.3 shows the spread of responses by (self-selected) sector. It had been hoped to recruit more from the Life Sciences and Creative sectors, but it proved necessary to pool over headings to get groups large enough for cross-tabulations. The largest group of respondents classified their firms as 'Services', with 'Financial' coming second. There were insufficient 'Food & Drink' to separate them from 'Retail'. Another 12 respondents represented some form of 'Government'. By combining 'Tourism' with 'Ferries' a group of 10 was obtained. Similarly, by combining 'Energy' with 'Forestry' gave obtained a group of 8. Other sectors were straightforward.
|SECTOR DESCRIPTION||NUMBER OF RESPONDENTS||CODE|
|Digital, Creative, Telecoms||9||D|
|Energy and Forestry||8||E|
|Financial and Business Services||18||F|
|Life Sciences, Medical, Veterinary, Academic and Training||7||L|
|Food, Drink and General Retail||12||R|
|Tourism and Ferries||10||T|
The results by considering impressions of reliability change are considered first. The general order will be to cross-tabulate by location first and then by company sector. Table 5.3.4 shows the overall result, and the results by location. Overall, 41% of respondents feel that roads have become less reliable, with an equal number noticing no difference. 15% thought roads were becoming more reliable, and 5% expressed no opinion. Trying to interpret those figures by looking by area, it was seen that those in the north (Aberdeen and Inverness areas) drove that result, along with the 'Other' location category. For the bulk of Scotland, only about 20% to 30% felt that reliability was getting worse, with 22% feeling that roads were getting more reliable. That last figure is, though, heavily driven by the 48% reporting that in the Glasgow area.
|SECTOR||LESS RELIABLE||THE SAME||MORE RELIABLE||DON'T KNOW||TOTAL RESPONSES|
Table 5.3.5 looks at reliability as seen by the different sectors. The Tourism/Ferries sector reports no deterioration in reliability. For Manufacturing, also, the percentage saying more reliable outweighs those saying less reliable. The Financial sector, on the other hand, is heavily of the view that reliability has got worse.
Table 5.3.6 reports responses regarding actual impacts on companies of trunk road unreliability in the last year. The only impact to have affected more than half of respondents was difficulty with staff travel. By contrast, only 10% of respondents reported any difficulty in attracting customers in this regard. Roughly 30% of respondents reported the remaining four impacts (Reduced productivity/Sales; Additional Transport Costs; Delays to Time-Critical Deliveries; and Additional Staff Costs).
|SECTOR||LESS RELIABLE||THE SAME||MORE RELIABLE||DON'T KNOW||TOTAL RESPONSES|
|D – Digital etc.||38%||38%||25%||0%||8|
|E – Energy/Forestry||56%||22%||22%||0%||9|
|F – Financial||88%||13%||0%||0%||8|
|R – Retail/Food/Drink||28%||44%||11%||17%||18|
|L – Life Sciences etc.||42%||58%||0%||0%||12|
|T – Tourism/Ferries||0%||75%||25%||0%||8|
|C – Construction||50%||50%||0%||0%||8|
|M – Manufacturing||25%||25%||33%||17%||12|
|S – Services||52%||30%||17%||0%||23|
|G – Government||30%||60%||10%||0%||10|
|Additional Transport Costs||75||38||3|
|Delays to time-critical Deliveries||86||27||3|
|Additional Staff Costs||81||33||2|
|Difficulty in Attracting Customers||100||12||4|
|Difficulties with Staff Travel (on business, and commuting)||46||69||1|
Respondents were asked if there were particular causes for the trunk road unreliability affecting their company, but there were not many responses and few that came up more than once (indicated here with grouped frequency in brackets). Reasons mentioned were:
- Badly planned road works (2);
- Lack of motorways north of Glasgow;
- Inadequate planning for weather/accidents/flooding (3);
- Lack of maintenance, potholes, drainage (3);
- Lack of investment (3);
- Unnecessarily low speed limits;
- Badly timed traffic lights.
Respondents were also asked to give examples of a journey where there had been a problem with unreliability, and that elicited a wider range of concerns, but expressed in a way very specific to the journey in question, and therefore not suitable for listing (egg. the closure of a particular road following a fatality; and "Amounts of Traffic at 1500 on a Thursday during the school holidays"). Table 5.3.7, however, reports average ratings of the scale of various impacts on trunk road reliability, together with a break down by location.
As can be seen from Table 5.3.7, the chief impacts on reliability are felt to be the concentration of traffic at particular times and roads being insufficiently large to cope. Road works came in third place, with the remainder only being awarded "half marks". By location, road design received the highest rating in the Inverness area. That area, ignoring 'Others', also gave the highest rating to road size; with Glasgow giving it the lowest rating.
Concentration of traffic at particular times was felt to be particularly impactful in the Aberdeen area, but not a great worry in the South West and the Inverness area. Road works were particularly blamed in Edinburgh and Tayside (and Others). The weather did not appear to have been much of a problem, the previous 12 months having been unusually clement in Scotland. Accidents were felt to have particular impact in the Inverness area, but received in Edinburgh the lowest rating in the whole table.
|Roads not large enough||7.6||6.0||7.1||7.0||5.6||8.1||6.8||8.8||7.1|
|Traffic concentrated at particular times||8.4||5.6||7.6||7.5||7.0||5.4||7.7||7.3||7.1|
|Roads not large enough||7.5||5.7||9.3||5.8||6.3||6.8||9.4||7.3||7.5||7.3|
|Traffic concentrated at particular times||7.6||5.8||7.5||7.6||7.8||6.8||8.1||6.1||7.4||6.4|
Ratings of impacts by sector are shown in Table 5.3.8. Starting with Construction (C), they highly rate traffic concentration and road size, but not road design. Digital etc. (D) rate road works highest, but not weather. Energy/Forestry (E) rate size of roads very highly, but accidents lowly. Financial (F) rate traffic concentration highest, and accidents lowest. Government (G) also rate traffic concentration highest, but with both road design and accidents lowly rated. Life Sciences etc. (L) rate everything very closely, except a low rating for weather. Manufacturing (M) give the highest rating in the whole table, and that goes to road size. Conversely, they give accidents and weather particularly low ratings. Retail, Food & Drink (R) rate road size highest, and accidents lowest. Services (S) rate road size and traffic concentration highest, with accidents and road design lowest. For Tourism (T) it is road works and road size that have the largest impact, with road design rated lowly.
Table 5.3.9 gives some insight into respondents' need for access to certain groups of people. Asked to provide a rating going up to 10 for most important, respondents declared that access to Customers came highest with a median rating of 8, and mean rating of 6.7. Access to a skilled work force was close behind with median 7 and mean 6.0, whilst access to suppliers was deemed relatively unimportant with a median of just 5 and a mean of 5.3.
|IMPORTANCE FOR BUSINESS||MEDIAN RATING OUT OF A MAXIMUM OF 10||MEAN RATING OUT OF 10|
|Access to Customers||8||6.7|
|Access to Suppliers||5||5.3|
|Access to Skilled Work Force||7||6.0|
Table 5.3.10 disaggregates these importance ratings by location. Naturally, the variations in the figures will be largely driven by the range of industries forming the sample in the various areas. Access to customers is most important in the Tayside and Inverness areas, and relatively unimportant in Edinburgh, the South West and Central areas. Access to suppliers was most important in the Inverness and Aberdeen areas, but very lowly rated in the South West. Access to a skilled workforce was most important in the Aberdeen area, and least important in the South West.
|Access to Customers||7.1||5.3||5.1||5.5||6.2||8.2||8.6||8.2||6.7|
|Access to Suppliers||6.4||2.3||5.0||4.7||5.1||6.9||5.0||5.7||5.3|
|Access to Skilled Work Force||7.1||2.7||5.6||5.5||6.4||6.3||5.6||6.8||6.0|
Table 5.3.11 gets to grips with the differences by sector. Access to customers was most important in the Life Sciences etc. sector and the Tourism/Ferries sector. It was also important in the Digital etc. and Services sectors. Those four sectors do seem the most likely to need access to customers. Possibly surprisingly, though, it was least important for the Manufacturing sector. Access to suppliers was most important in the Retail/Food&Drink sector, which seems sensible. It was least important for the Life Sciences etc. sector, which again seems sensible. Other sectors expressing low importance for access to suppliers were: Services; Government; and Financial. Access to a skilled workforce was most important for the Energy/Forestry sector, and also relatively important for the Digital etc. and Financial sectors. It was least important for the retail etc. and Tourism/Ferries sectors.
|Access to Customers||6.6||7.4||5.9||5.9||6.0||8.4||4.5||6.8||7.3||8.1|
|Access to Suppliers||6.1||5.6||6.7||4.9||4.6||3.1||5.4||7.4||4.4||6.7|
|Access to Skilled Work Force||4.9||7.3||8.4||7.0||6.2||5.1||6.3||4.4||5.9||4.2|
Table 5.3.12 provides added detail on why firms would like to see Scottish trunk road reliability improved. Respondents were asked to choose the 5 most important impacts, from the list of 9 shown in the table, and rank them 1 (for most important) to 5 (for least important). There were many non-responses, and lots of cases where the response was not the expected 5 digits 1 to 5 spread over the 9 columns. It was virtually the last question and fatigue may have set in. Making the best of what there is, by cleaning and averaging the responses, the table presents a composite ranking. In that ranking, 1 denotes the impact the respondents seemed to feel was most important, down to 9 for least important. The two most important appear to be "Access to Customers" and Productivity, in that order. Not far behind were Transport Costs. Then there is a big gap back to "Attractiveness of area" in 4th place, followed by "Staff Recruitment" and "Access to Suppliers". By far the worst ranked was "Exports" though, as that was the last in the list presented, it may just be that respondents had used their 1 to 5 by then.
|Attractiveness of area||4|
|Access to Suppliers||6|
|Access to Customers||1|
|Links between firm's locations||7|
Table 5.3.13 provides detail by location. Here the individual rankings from 1 down to 5 have been averaged, together with '6' used for all unranked reasons. A low score, therefore, indicates that the stated reason has been deemed important. Improving the attractiveness of the area is felt to be important in Glasgow and Tayside and, to a lesser extent in the Aberdeen, Edinburgh and Inverness areas. Enhancing productivity was a very important reason in the Aberdeen area, and was important everywhere except Tayside. Better access to suppliers was an important reason in Glasgow, and borderline important in the Aberdeen and Inverness areas. It was judged unimportant in the Tayside and Central areas. Better staff recruitment was felt an important reason in Glasgow, and borderline important in the Aberdeen, Central and Edinburgh areas. It was felt unimportant in the Inverness area. Better access to customers was an extremely important reason in Tayside, and very important in the Central and (to a lesser extent) South West areas. It was important in all other areas. Improved transport costs were important in all areas. Linking firms' locations was only important in the Central area. It was deemed unimportant in the South West and Inverness areas. Improving business confidence was only important in the South West. Helping with exports was nowhere important, and in the South West was awarded the 'perfect 6', i.e. totally unimportant.
|Attractiveness of area||3.8||4.7||4.3||3.8||3.5||3.9||3.6||4.4||3.9|
|Access to Suppliers||3.9||4.5||5.1||4.2||3.6||3.9||5.0||4.2||4.1|
|Access to Customers||3.9||2.8||2.1||3.4||3.2||3.3||1.4||3.2||3.2|
|Links between firm's locations||4.4||5.2||3.7||4.0||4.2||5.2||4.4||3.4||4.2|
Table 5.3.14 looks at the responses by sector. Improving the attractiveness of the area was important for the Financial sector and Tourism/Ferries sector, and less so for the Digital etc., Life Sciences etc., and Manufacturing sectors. Improving productivity was a very important reason in the Energy/Forestry and Services sectors, and was important in all other sectors than Tourism/Ferries. Improved access to suppliers was important in the Creative, Digital and Retail sectors. Better staff recruitment was a very important reason in the Manufacturing sector, and important for the Financial sector. Improved access to customers was particularly important in the Life Sciences etc. and Retail sectors, and also very important in most other sectors. Improved transport costs were a very important reason in the Construction and retail sectors, and important in most other sectors. Links between firms' locations were never judged as an important reason. Improving business confidence was very important in the Life Sciences etc. sector, and important in the Financial sector. Finally, help for exports was never judged an important reason.
|Attractiveness of area||4.6||3.7||4.9||3.1||4.2||3.7||3.7||4.1||4.1||3.3|
|Access to Suppliers||3.4||3.3||3.9||4.5||5.2||4.3||4.7||3.3||4.0||4.3|
|Access to Customers||3.0||3.0||4.9||3.5||3.2||2.3||4.0||2.3||2.9||3.1|
|Links between firm's locations||3.9||4.6||4.6||4.1||4.4||4.7||4.2||3.9||4.2||4.2|
Finally, the responses to two interesting questions are reported. Firstly, respondents were asked for a "guesstimate" of the value to their company from improvements that would make the road network totally predictable and reliable. This was envisaged as giving an upper bound on the value of reliability to users. Only 23 respondents felt able to provide a numerical estimate, of which 6 replied zero. The median response was £3000, which seems plausible, while the mean (£485,000) was clearly swayed by two responses of £5 million.
Secondly, respondents were asked to estimate the travel time for one day in 5, where the remaining 4 were all "as now", that would make that set of 5 journeys equally desirable to a set of 5 days where travel time was always 10 minutes longer than now. This is a somewhat complex question, and it was pleasing when the pilot respondents were able to provide responses. Essentially, there are two situations: in the first there is travel time variability; in the second there is no travel time variability but 4 out of 5 journeys will take longer than now (i.e. currently). The Median response was 30 minutes, and the Mean response was 35 minutes. It was preferred to work with the median as it excludes outliers, who may not have properly understood the question.
Denote the current travel time as T, the value of a minute of standard deviation of travel times by VOR, standing for value of reliability, and the value of a minute of travel time as VOT. The standard deviation of option 2 is zero, as travel time is (T+10) minutes each day. Then denote the standard deviation of option 1 by S, and the travel time reported for the 5th day by (T+X). Working in minutes gives:
|OPTION 1: (T, T, T, T, T+X)||TIME1 = 5T+X||SD1 = S|
|OPTION 2: (T+10, T+10, T+10, T+10, T+10)||TIME2 = 5T+50||SD2 = 0|
It is known, by design, that the two options are equally valued, so we have per day:
(VOT)(T+X/5) + (VOR)(S) = (VOT)(T+10) + (VOR)(0)
whence (VOT)(10 – X/5) = (VOR)S
The reliability ratio, RR, was defined in Section 4 as
RR = (VOR)/(VOT)
which here gives
RR = S/(10 – 0.2X)
Essentially, some extra travel time, (10 – 0.2X), is being accepted in return for avoiding the unreliability, measured by standard deviation, resulting from the travel time being X minutes one time in 5 and zero otherwise. If the unreliability was zero valued, respondents would just judge on travel times, and report an X value of 50 minutes. RR would then be zero. Both options would take 50 minutes for the 5 days, and it would not matter how the travel times were distributed over the 5 days. Hence, reported X values should be below 50. There were actually 16 above 50, those respondents presumably having misunderstood the question. By working with the median, these large values will just be treated as a value above the median (and the magnitude ignored).
The median response was X = 30, so S = sd(T, T, T, T, T+30) = 13.42,
RR = (10 – 6)/13.42 = 0.3
This value is at the bottom end of values found in the literature, suggesting that reliability is valued less likely by the company respondents in Scotland than by the respondents to earlier surveys.