66798 China Transport Topics No. 04 February 2012 High-Speed Rail – The First Three Years: Taking the Pulse of China’s Emerging Program Richard Bullock, Andrew Salzberg, and Ying Jin World Bank Office, Beijing High-speed rail services have now been operating in China for three years. How are they performing? What has happened to the conventional services they parallel? What has been the impact on the airlines? Little comprehensive information has been published to date but a general picture is emerging in which high-speed rail, as in other countries, is competing strongly on short and medium-distance routes up to 1,000 km while air remains dominant over longer distances. Overall, however, diverted air passengers have not been a major source of high speed rail ridership. A larger source has been ‘generated’ trips: new trips by passengers who were induced to travel by the greater convenience of high speed service. Based on this evidence and the continuing strong growth in Chinese urban populations and incomes, we are cautiously optimistic about the long-term ridership (and hence economic viability) of the major trunk railways of the high-speed rail network in China. This optimism is tempered by the need to develop a sustainable financing mechanism in the short to medium term and to carefully weigh the costs and benefits of the peripheral extensions of the network. THE NETWORK SO FAR Most lines have a large proportion of tunnel and China began to operate network-wide 200 km/h viaduct (in hilly areas)2 or elevated structure (in services in April 2007 as part of what is known as flatter areas)3, the latter to conserve farmland the Sixth Speed-Up1 but the first high-speed rail and minimize severance. The construction cost (HSR) service on a dedicated line was the Beijing naturally depends on the proportion of such – Tianjin service opened in August 2008 just tunnels and structures but typically ranges from prior to the Olympics. Since then, the network RMB 80-120 million per km (US$13-20 million) has grown to over 6,800 km, with a total of excluding stations. Station costs naturally 16,000 km to be completed as part of the long depend on their frequency and scale of range plan that runs to 2020 (Figure 1). construction but can add up to a further 30% for the shorter lines serving major cities. HSR CONSTRUCTION STANDARDS AND COSTS The new HSR lines are of two general types: HSR SERVICES AND FARES trunk lines, which are passenger-only lines The lines generally operate reasonably designed to operate at a maximum speed of 350 intensively, typically an hourly or half-hourly km/h, and secondary mainlines and regional service between 7am and midnight (Table 1), intercity lines designed with a maximum speed normally with 8-car sets. The main exceptions of 250 km/h, some of which are passenger-only and others which can also carry express freight 2 The 835 km of new construction for the Guiyang - such as container services. Guangzhou railway has 54% of its length in tunnel and 29% on viaduct and bridges. 1 3 A trial 200 km/h section was opened in 2002 between The 357 km Shijuazhuang – Zhengzhou section of the Qinhuangdao and Shenyang which was upgraded to 250 Beijing – Guangzhou north-south trunk line has 69% of its km/h in 2007. These ‘speeded-up’ services, together with length on viaduct and bridges. the 250km/h and 350km/h HSR services, are collectively known as the CRH (China Railway High-speed) services. China Transport Topics No. 04 2 World Bank Office, Beijing Figure 1 Chinese High Speed Rail (HSR) Network as of February, 2012 are lines that have been opened in advance of In 2007, CRH services totalled about 40 million adjacent sections of the network (such as train-km per year, all on conventional ‘speeded- Zhengzhou–Xian) and so are currently only up’ lines. By 2011 this had increased to over 250 providing service for local trips. million train-km per year, of which 85% was on dedicated lines with a maximum speed of 250 Table 1 Selected Service Patterns (as of July km/h or more. This represents about a quarter 2011) of the total passenger train-km operated on the First Last Services network. Section dep arr /day(1) Wuhan – Guangzhou 0700 2349 50 There are national fare scales for the two speeds Zhengzhou – Xian 0800 2246 12 but in practice HSR fares vary slightly from line Changchun – Jilin 0656 2124 28 to line (Table 2). Beijing – Tianjin 0630 2333 76 Beijing – Shanghai 0700 2327 49 Wenzhou – Fuzhou 0734 2245 27 (1) Per direction; some lines also have services to/from intermediate points China Transport Topics No. 04 3 World Bank Office, Beijing Table 2 Economy HSR Fares (as of November increased to 23 trains and 47 trains respectively. 2011) The distinction is relevant as passengers wishing to travel relatively short distances (for example, Economy fare from Changsha to Guangzhou) are generally less Section RMB RMB/km able to purchase a ticket on a ‘through’ train 350km/hr design speed 4 than a ‘local’ train7. Following the opening of the Wuhan – Guangzhou 465 0.48 HSR line, local trains were reduced to only 6 services, with the through trains increased to 52 Zhengzhou – Xian 230 0.46 services (see table 3). Beijing – Tianjin 55 0.46 Beijing – Shanghai 555 0.42 Table 3 Southbound conventional services into Guangzhou 250km/hr design speed Wenzhou – Fuzhou 91 0.31 Section 2007 2009 2011 Changchun – Jilin 32 0.29 ‘Local’ Taiyuan – Beijing 149 0.29 Wuhan-Guangzhou 6 9 2 Changsha-Guangzhou 12 15 3 For comparison, air fares are typically around RMB 0.70/km, with shorter trips a bit more Shaoguan-Guangzhou 17 23 6 expensive, and bus trips are about RMB ‘Through’ 0.35/km, with sleeper buses slightly more Wuhan-Guangzhou 21 24 25 expensive. The lower-speed lines are thus price- Changsha-Guangzhou 25 30 31 competitive with bus while the higher-speed Shaoguan-Guangzhou 42 47 52 lines are generally cheaper than air except where heavily discounted air fares are offered.5 Table 4 shows comparable figures for three other corridors, showing a general reduction in EFFECTS ON CONVENTIONAL RAIL OPERATIONS local trains and some adjustments to through trains in response to the additional capacity. Service reductions on conventional lines paralleling new high speed services have in some Table 4 Service changes – other corridors case been extensive. In the case of the Wuhan – Section 2007 2009 2011 Guangzhou service, in 2007 there were 17 ‘Local’ services daily to Guangzhou originating within the Wuhan – Guangzhou corridor (termed ‘local’ Zhengzhou – Xian 4 6 0 trains for the purposes of this discussion), with a Shijiazhuang -Taiyuan 1 4 4 further 42 trains from origins outside the Changchun – Jilin 0 6 2 corridor (termed ‘through’ trains for the ‘Through’ purposes of this discussion)6. By 2009, these had Zhengzhou – Xian 63 58 57 4 On opening, the average operating speed of non-stop Taiyuan – Beijing 11 9 238 services on these lines reached as high as 90% of these Changchun – Jilin 11 13 10 design speeds. Operating speeds have recently been temporarily reduced by about 12–15 %, apparently to Shaoguan etc) and terminating in Guangzhou would be conserve energy and reduce vehicle maintenance costs. 5 classified as ‘local’. For reference, rail fares on conventional lines range from 7 MOR’s ticket quota system generally favors long distance RMB 0.11/km for the slowest services to nearly RMB travel at the expense of short-distance trips to maximize 0.30/km for the 200km/h services (almost equal to the cost network utilization. per km on new 250km/h dedicated track). 8 6 Includes several long-distance trains (e.g. Beijing – For example, a train originating in Beijing and terminating Wulumuqi (Urumqi) and Beijing - Xian) rerouted to take in Guangzhou would be classified as a ‘through’ train in this advantage of the increased corridor capacity. corridor, while a train originating in Wuhan (or Changsha, China Transport Topics No. 04 4 World Bank Office, Beijing In some corridors, many of the conventional Figure 2 Rail passenger demand 2002-10 services removed have been overnight express trains. For example, between Changsha and Guangzhou in 2009, 12 ‘local’ trains departed Changsha between 7 PM and midnight. With an 8 to 9 hour travel time, these overnight trains offered a mix of low cost and convenience since on-board hours could be spent sleeping. In 2011, after high speed opening, only 3 such trains departed during the same time period. The removal of these services has been reported in the media as a source of irritation among some passengers. Another important impact of these service The first dedicated high speed line to open, from changes is freed-up freight capacity. For Beijing to Tianjin in 2008 with a relatively short example, the 12 conventional passenger services length of 117km, is currently carrying about 25 removed from the Wuhan-Guangzhou line will million passengers per year. The line from leave capacity for approximately six additional Wuhan to Guangzhou (968 km) is carrying about freight services, providing important additional 22 million passengers per year but most of these revenue generating capacity to the Ministry of passengers do not travel the full length of the Railways. line. The average density of passenger traffic on this line is about 10 million passengers per year.9 RIDERSHIP The lowest ridership is probably found on the Detailed official ridership figures for the HSR Zhengzhou – Xian and Nanchang – Jiujiang services generally are not available but indirect services; both are currently operating as isolated evidence suggests ridership is broadly favourable sections of future trunk lines (Figure 1 shows compared to international experience. In 2010, these existing lines and their place in the future 290 million passengers (17% of the total carried network). As the rest of the network is in China) travelled on services operating at completed, these lines will be able to provide 200km/h (both on dedicated high speed lines service to a wider range of origins and and ‘speeded-up’ conventional lines as described destinations, helping to increase service above). The important contribution of these frequency and ridership. services to overall growth in rail passenger demand since 2006 is shown in Figure 2, where A key feature observed in HSR demand to date the incremental growth since 2007 is roughly has been the high level of generated trips, i.e. equivalent to the volume currently carried on trips made by those who have been induced to 200km/h plus services. travel by the improved service levels (speed, frequency, reliability, and comfort) of HSR. In the 9 The average density is derived by dividing the passenger- km by the length of the line in km. For the short Beijing- Tianjin line, where almost all traffic travels end to end, the density is approximately equal to the number of passengers, 25 million. For comparison, the Tokyo – Osaka line carried 141 million passengers in 2010, with an average density of 85 million. The corresponding figures for the Taiwan HSR are 37 million passengers with an average density of 22 million. China Transport Topics No. 04 5 World Bank Office, Beijing case of the Wuhan - Guangzhou line, in 2009 dramatic indication of the latent demand for the prior to the opening of the new high speed line, type and quality of service being offered by HSR. the corridor was carrying an estimated 45 million local rail passengers (i.e. those with both origins In the case of the line from Changchun to Jilin, a and destinations within the corridor). In 2010, 111 km HSR line with a design speed of 250 the first full year of operation of the new HSR km/hr, an estimated 4 million passengers per line, the estimated total travel on both the year previously travelled between the two cities conventional and HSR lines was about 55 million, on conventional rail services. The new high of which 20 million are travelling on HSR.10, 11 Of speed line, which opened in January 2011, is these, about 1 million appear to have come from estimated to be carrying around 10 million competing air services (see next section) and passengers per year. Based on information from about 10 million have transferred from intercity bus operators, an estimated 2 million conventional rail services. A few have also per year have transferred from bus service to the transferred from bus and car. Based on these new HSR line. An estimated 1 million per year estimates, of the 20 million travelling on this HSR have transferred from local rail, leaving roughly line each year, about 50% have transferred from 7 million per year of the HSR passengers who conventional rail, about 5% have transferred have either transferred from private vehicles or from air and 45% have been either generated or been generated, for an estimated generation transferred from bus and car, with the rate of 40-50%. overwhelming majority of these being generated. Despite the high rates of generation noted Put another way, for the Wuhan - Guangzhou above, patronage on some of the lines remains rail corridor, about 25% of the conventional rail substantially below the opening-year forecasts passengers have transferred to HSR while the developed in their respective feasibility studies. overall rail market (conventional plus HSR) has increased by about 20%. One reason for this is that many feasibility studies assumed a wholesale transfer of The Beijing to Tianjin service shows a similar conventional rail passengers to HSR on opening. pattern. Prior to 2008, about 8 million Few studies made a realistic allowance for HSR passengers travelled on the conventional rail fares being much higher than fares on service each year but this reduced by about 50 conventional rail (as described above), and many percent following the introduction of the HSR made a simple assumption that all existing services. The bus service is also estimated to conventional passengers would transfer to HSR. have lost about 1 million passengers to HSR. 25 In fact, as mentioned above, using the Wuhan to million passengers per year are now using the Guangzhou line as an example, only an new HSR line. On this basis, roughly 20 million estimated 25 per cent have made the transfer. passengers per year of the current demand This transfer rate will, however, increase as either transferred from private vehicles incomes increase, providing a source of long- (including minibus) or has been generated. As in term growth for high speed rail ridership. Wuhan, the overwhelming majority of these have probably been generated trips. This Another important reason for the discrepancy suggests a conservatively estimated rate of between forecast and actual ridership is that generation of 65% or greater on this line, a many feasibility studies assumed that the entire 2020 network would be in place by the opening 10 The Wuhan-Guangzhou high speed line opened on year of any individual line. In practice, as shown December 26, 2009. in Figure 1, this final network configuration is still 11 Based on ridership growth trends in the corridor, the annual growth in demand would have been about 1.5 some years off for most corridors and demand million trips in the absence of the new line. will likely build up gradually as the network is China Transport Topics No. 04 6 World Bank Office, Beijing completed, making the ‘ramp up’ more marked Figure 3 Changsha – Guangzhou air services than usual on these lines. For example, the feasibility study for the line from Zhengzhou to Xian estimated that local trips between origins and destinations within the corridor would represent just 20 percent of the total passenger- km on the line. The other 80 percent was represented by longer distance HSR trips – i.e. between destinations beyond either Zhengzhou or Xian – that are not yet possible with the current network configuration. As the rest of the network is completed, travel between these other origin-destination pairs will A similar pattern can be seen for the Wuhan – become possible by high speed rail, supporting Guangzhou route (a distance of approximately ridership growth on the Zhengzhou-Xian line. A 900km) (Figure 4), although in somewhat less similarly positive ‘network effect’ will be felt by dramatic fashion. Passenger demand has fallen other lines as the remaining pieces of the by about one-half, from about 120,000/month to network enter into operation in the coming 60,000 per month, while flights have reduced years. from 900 per month to about 600 per month. Figure 4 Wuhan – Guangzhou air services IMPACT ON OTHER MODES For distances up to approximately 500 km, HSR has had a strong impact on air demand. As has been widely reported, some short-distance air services have been completely withdrawn after an HSR line has opened; air routes from Zhengzhou to Xian and from Wuhan to Nanjing both survived only a few months after the opening of HSR. Figure 3 shows monthly air services (sum of both directions) between Changsha and Guangzhou, a distance of approximately 600km. Passenger demand has reduced from about 90,000 per month prior to However, this impact appears to die away quite the line opening to about 30,000 at the present quickly for distances over 1,000 km. The opening time and services have reduced from 750 per of the Beijing – Shanghai line (approximately month to 250 per month. 1,300 km) has had very limited impact on air travel between the two cities. In contrast to longer routes where the chief competition in terms of travel speed and cost is air, the main competitors for shorter intercity routes are bus and private vehicles (car and minibus). Systematic volume data for both of these modes is unfortunately not often available but, based on evidence collected from bus service operators, they have often been hard-hit. The competing bus service along the Changchun China Transport Topics No. 04 7 World Bank Office, Beijing to Jilin route described above charges roughly rescheduled in one way or another for all but the the same fare as the new HSR line but offers a busiest Chinese lines. Conventional solutions to much lower quality of service, and has thus been this issue include extending the tenor of the all but eliminated; the service has reduced from loans or sculpting debt repayments to better a bus every 5-10 minutes to one or two buses a match the growth in demand over time. Other day, travelling via intermediate towns. The options include combining lines into a few large competing bus services along the Beijing to groups so that the main lines can support their Tianjin route still maintain a significant price feeder branches, or providing some funds from advantage over HSR (about half the HSR fare) the conventional railway to reflect the benefit of but nonetheless have not been able to hold onto the additional capacity which is being made to their riders, having lost over half their available for them. patronage. SUMMARY FINANCIAL PERFORMANCE It is now three years since the first high-speed Very little information is publicly available on the rail line in China opened. The total volume financial performance of the HSR network. There carried is already larger than that on the French seems little doubt, based on experience TGV services and is rivalling the volume on the elsewhere and evidence collected in China, that Japanese Shinkansen services. It will continue to most lines, except possibly isolated lines like grow rapidly as the many lines under Zhengzhou to Xian and Jiujiang to Nanchang, are construction are completed and as urban covering their immediate cash costs (rolling incomes and population in China continue to stock and infrastructure operations and rise. maintenance costs). The next important financial hurdle for any railway is its ability to cover its Ultimately, the appropriate time frame to deliver interest payments on debt, and it is likely that final judgement on the program will be some of the better-performing services (with measured in decades, not years. However, densities of 10-15 million passengers) are evidence accumulated to date provides some already doing that. key insights. Of course, covering interest payments is not the This evidence suggests that HSR can compete same thing as making a profit, as depreciation on effectively with air at distances up to 1,000 the capital investment is a substantial cost. kilometres but that it can gain only a limited However, most of the depreciation is for share over longer distances. Over shorter infrastructure with a very long life. Given the distances, it seems able to take almost all the continuing growth of urban populations, bus market as long as its stations are incomes, and the overall network, ridership (and conveniently located. However, gains from both hence revenue) growth should continue to ramp of these markets have been dwarfed by the size up over time. If, in accounting terms, of the new, generated market for rail trips, depreciation is deferred now, there should be representing passengers who had not travelled ample time to close this gap over time for lines in the corridor before. We estimate that are reasonably well performing. approximately half or more of HSR passengers appear to be these new travellers. The final financial threshold for a railway is its ability to repay the principal on its loan. Only a This finding is significant, pointing to a large handful of HSR lines in the world have been able latent demand for the quality of service provided to reach this milestone. It seems likely that by HSR. These generated trips also suggest that repayment of principal will ultimately need to be businesses and individuals are already modifying China Transport Topics No. 04 8 World Bank Office, Beijing their behaviour to take advantage of this new (jscales@worldbank.org) or Gerald Ollivier mode of transportation, providing some (gollivier@worldbank.org), from the Beijing preliminary evidence of the wider economic World Bank Office. benefits to be provided by these lines in the long term. Based on these trends and the continuing Any findings, interpretations and conclusions strong growth in Chinese urban populations and expressed herein are those of the authors and do incomes, we are cautiously optimistic about the not necessarily reflect the views of the World long-term ridership (and hence economic Bank. Neither the World Bank nor the authors viability) of the major trunk railways of the HSR guarantee the accuracy of any data or other program in China. information contained in this document and accept no responsibility whatsoever for any This optimism is tempered by some significant consequence of their use. obstacles that will need to be overcome in the short to medium term. Little data is publicly available on the finances of the new lines but, based on experience elsewhere, even well- performing railways capable of covering their cash running costs and interest on their debt will almost certainly be unable to repay the principal without some long-term financing arrangements. Given the large scale of the HSR investment program, this issue will need to be addressed in the short term. The recent slowdown in construction activity may also provide an ideal opportunity to review the as yet unbuilt peripheral sections of the network, ensuring that they will not create an undue financial burden on the overall network. ********** Richard Bullock is a transport consultant specialising in railway business planning, financial modelling and project evaluation. Andrew Salzberg is a transport consultant working with the World Bank and specializing in transport and urban development planning. Ying Jin is Deputy Director of Martin Centre for Architectural and Urban Studies, Cambridge University and Fellow of Robinson College, Cambridge. This note is part of the China Transport Note Series to share experience about the transformation of the Chinese transport sector. For comments, please contact John Scales