Topic 33 FREIGHT in GREAT BRITAIN – MODAL SPLIT

1. Data in this note refers to the year 2010 except where stated. 
2. RTS denotes Road Traffic Stats
3. FS denotes Freight Stats
4. CSRGT denotes the Continuous Survey of Roads Goods Traffic.

Summary

Table 0401 of the 2011 TSGB provides the following estimates of Tonne-km: 151bn by road, 19bn by rail, 42bn by water (mainly inshore shipping) and 10bn by pipe line, where the numbers relate to Great Britain.  Those yield 68% by road, 8.5% by rail, 19% by water and 4.5% by pipe or if water and pipe are ignored, 88.8% by road and 11.2% by rail. 

This note shows that, subject to the caveat below, the road freight in that table should be increased by 24%.  The effect on modal split is to provide 186.6bn tonne-km or 72.4% by road leaving 7.4% by rail, 16.3% by water and 3.9% by pipe. Or, if water and pipe are ignored, 91% by road and 9% by rail. 

Further, we now estimate that 73% of the nation’s road freight is on the Motorway and Trunk Road system, compared with previous estimates, based on the Heavy Goods Vehicle veh-km, of 64%.

These estimates include 12bn Tonne-km in light goods vehicles.  (This value is the difference between 151bn Tonne-km in TSGB Table 0401, where the effect of light goods vehicles is included, and the value of 139bn Tonne-km for domestic road freight available from other FS tables where freight in light goods vehicles is not included).

A caveat: 

Possibly the vehicle-km missing from the estimate made from the CRRGS are not randomly distribute but are concentrated on empty running.  If so then our estimate for Tonne-km will be too high.

The Problem 

1. That data in TSGB 0401 makes an allowance for light goods vehicles (vehicles of less than 3.5 tonnes).  Without the latter there would be 139bn Tonne-km by road, freight Table RFS0107.  However, both tables omit international freight in foreign and Northern Ireland registered lorries.  That at least should be added if the data is to be compared to other modes.  It amounted to 9.1bn tonne-km in 2010, see data here.
2. Separately from that we had an interest in calculating the proportion of road freight carried by the strategic road network.  To that end we requested, from the freight stats team, the vehicle-km and tonne-km by class of vehicle, defined by numbers of axles.  That data would enable us to calculate average loads for the various axle configurations.  Multiplying by the corresponding vehicle-km, available by special request from the Road stats team, would provide the tonne-km.
3. The immediate problem was that the vehicle-km provide by freight stats differed from those provided by Road stats as set out below:  Detail is in the working spread sheet here.

4. The differences are so large as to undermine confidence in the statistics as a whole. We have 40% more vehicle-km from Road stats than from Freight stats, let alone the vast differences at vehicle class level.
5. We informed the DfT statistics people of the problem in September 2012.  On 25^th January 2013 we received notification of the Discrepancies Report.

The discrepancies Report.

This report says that The RTS data (a) overestimates 2-axle Goods vehicle traffic by 1,400 million vehicle-km (b) includes heavy vehicles that are not classed as Goods Vehicles by the FS survey, but makes no estimate of the effect (c) Classifies vehicles travelling with one axle raised as though from the lower axle class but again makes no estimate of the effect. 

Additionally the report says that the FS data underestimates traffic attributable to rigid bodied goods vehicles by 20% and that attributable to artics by 11%.  Further, as above, the CSRGT does not capture foreign and Northern Ireland lorries .  These vehicles are however inevitably included in the RTS data.  The report says the foreign and Northern Ireland lorries account for some 1035 million vehicle-km, mostly 5-axle artics.

The report is silent as to adjustments needed to the tonne-km estimates. 

More detailed comments are here.

Our adjustments

Calculations and sources providing our estimates of road freight and the modal split are as follows:

We started with the data at left in the first sheet in the spread sheet attached.  It provides tonne-km and vehicle-km by class of vehicle.  The source is the DfT''s  Freight Stats team.  We then combined vehicle types so as to correspond to those recorded by the DfT's Road Traffic stats team. That data is at right on the first sheet of the spread sheet.  Dividing the Tonne-km by the vehicle-km provided average loads for the RTS classifications.

The vehicle flows, as adjusted in Appendix F of the discrepancies report, are at top of the second tab of the spread sheet. We adjusted the RTS values by:

1. Reducing vkms to take account of non-goods vehicle HGVs included in the RTS values:

The RTS HGV vkms include HGVs not licensed as goods vehicles. The Discrepancies report says that these are mostly 2-axle rigids or 5-axle artics.  They include emergency services, farm tractors, mobile cranes and diggers along with road sweepers.  Table VEH 0102 provides 470,100 HGVS.  Table VEH 0103  provides 389,900 goods vehicles.  The difference is 80,200.  Our view is that all of them will have trivial annual mileages.   We assume 8,000 km per year per vehicle, providing 640m vkms.  We split that equally between the 2-axle and 5-axle artics, deducting 320 million vkms from each, on the assuption that they do not carry freight.

2. Adjusting the numbers in each class, but without changing the totals, to take account of the vehicles with one axle raised

The sub-classifications need adjusting to take account of the RTS classifying lorries with one axle raised as vehicles with one axle less than born.  If the percent running with an axle raised is p then the recorded veh-km for the highest axle class in the  RTS should be increased by dividing the recorded value by (1-p).  That should be subtracted from the class immediately below.  The difference in that flow class should then be divided by (1-p).  We set “p” to 35% for 6-axle artics, 10% for 5-axle artics, and to 5% for 4-axle and 3-axle rigids.  Readers may carry out sensitivity tests by varying any of those within the spread sheet.

Multiplying the adjusted RTS vehicle-km by the average loads from the FS data and summing yields our estimates of 174.6 Tonne-km for 2010.  Adding for freight in light goods vehicles provides 186.6bn Tonne-km

Applying similar principles to the HGV flows on the motorway and trunk road system provided an estimate of tonne-km on that network

Values for other years are estimated by apportioning the difference between our estimate of 186.6 bn tonne-km and the value of 151 bn tonne-km in TSGB 0401 according to the flow differences from Appendix A of the Discrepancies report. That produces the data below also available in the spread sheet at third tab.

Results

1. Our estimate of the road freight on roads in Great Britain for the year 2010 is 186.5bn tonne-km, an increase of 24% on the value of 151 bn tonne-km in TSGB table 0401
2. The effect on modal split is to provide 72.4% by road, 7.4% by rail, 16.3% by water and 3.9% by pipe. Or, if water and pipe are ignored, 91% by road and 9% by rail.
3. We estimate that the motorway and trunk road network carries 137bn tonne-km, including an allowance of 3,650m tonne-km for freight in light goods vehicles, (the latter calculated by apportioning the 12bn for all roads in proportion to light goods vehicle flows).  On that basis the motorway and trunk road system, carries 72% of the nation’s road freight.  In corroboration, the network carries 80% of the vehicle-km attributable to artics and nearly 70% of that attributable to 4-Axle rigids.

The following table summarises the results.

 The comparison of the RTS adjusted the CSRGT adjusted flows is fair.

The proportional calculation for other years provides the table below.

Sensitivity tests

Readers may a carry out their own sensitivity test by varying the proportions of RTS vehicles with one axle raised within the spread sheet, or altering other adjustments. 

As an alternative we have made estimates where the RTS values for total vehicle-km are taken as control totals but where the proportions in each vehicle class are adjusted to match those found in the FS data, see spread sheet , sheet 4.  That provided data immaterially different from the above.  E.g. an increase in Tonne-km with respect to the original CSRGT data of 25% instead of 24% and 71% of freight on the strategic road network instead of 72%

Conclusion

Subject to the caveat, cited above, namely, that possibly the vehicle-km missing from the estimate made from the CRRGS are not randomly distribute but are concentrated on empty running, significant, if not substantial, amendments are due to all, or nearly all, the road freight stats published by the DfT.

We also encourage the DfT to amend the road traffic stats vehicle classifications so as to report the axle classes as seen instead of classifying those with one axle raised as from the lower axle class