Home > Analysis, General > China Railways Safety Record: A Comparative Study

China Railways Safety Record: A Comparative Study

January 7th, 2013 Leave a comment Go to comments

Summary

  1. Railways are the safest means of transportation among all methods in this study.
  2. Railway safety-wise, China and France are about twice as safe as Japan and Germany.
  3. You are about 2.2 times as likely to die in an American airplane as in a Chinese rail car, traveling the same distance.
  4. You are about 200 times as likely to die in a vehicle on an American highway as in a Chinese rail car, traveling the same distance.

Background

China has experienced a building boom in railways, especially high-speed railways (HSRs).  There are many detractors outside of China pointing to its perceived shortcomings, especially its safety record, after the 2011 Wenzhou train crash that claimed 40 lives.  Recently the longest HSR line in the world between Beijing and Guangzhou became operational, and in reporting it almost all Western coverage brought up the Wenzhou crash.

The purpose of this post is to answer these questions: are Chinese railways safe, and how safe?

 

Methodology

To measure the safety of a means of transportation, in this post, the goal is to measure Deaths per Passenger-KM.  Chinese railways are being compared to their Japanese, French and German counterparts – plus American revenue-generating air travel (a subset of General Aviation) and American Highways.

Three caveats in this study:

  • I use only the available data for the most recent decade.  The primary reason is that the accident reporting prior to 2000 might be questionable.
  • The latest available passenger-km data for all countries is 2010 so the pasenger-km data is for 2001 – 2010.  However I use the fatality data for 2002 – 2011, because otherwise the Wenzhou crash won’t be included and it potentially will make China look better.  This overstates the fatality rate of China because China has a faster railroad passenger growth.
  • It covers all railroads (not just HSRs).  For the rationale, see this previous comment of mine.

 

Data

Passenger-KM (in thousands)

2001

2002

2003

2004

2005

2006

China/Railways [1]

463,660

480,310

456,000

551,196

583,320

635,327

Japan/Railways [1]

241,133

239,246

241,160

242,300

239,246

249,029

France/Railways [1]

71,209

73,227

71,937

74,014

78,306

78,465

Germany/Railways [1]

73,899

69,848

69,596

69,997

72,568

74,727

US/Air [2]

757,799

755,744

797,126

877,792

917,822

924,154

 

2007

2008

2009

2010

Total

China/Railways [1]

689,618

772,834

787,890

791,158

6,211,313

Japan/Railways [1]

252,579

255,865

253,555

244,235

2,458,348

France/Railways [1]

83,299

88,283

87,667

86,853

793,260

Germany/Railways [1]

74,740

76,997

76,772

78,582

737,726

US/Air [2]

952,549

913,491

866,645

906,595

8,669,718

Note:

1. http://data.worldbank.org/indicator/IS.RRS.PASG.KM/countries?display=default.

2. http://apps.bts.gov/xml/air_traffic/src/index.xml#CustomizeTable.  Select 1/2001 to 12/2010, Domestic, Total, Passenger, “Revenue Passenger Miles” and convert to passenger-KM (multiple by 1.6).

Fatalities

Year

2002

2003

2004

2005

2006

2007

2008

2009

China/Railways [3]

5

3

72

7

Japan/Railways [3]

112

France/Railways [3]

12

5

Germany/Railways [3]

6

21

US/Air [4]

33

64

65

40

50

19

70

50

 

Year

2010

2011

Total

China/Railways [3]

19

40

146

Japan/Railways [3]

112

France/Railways [3]

17

Germany/Railways [3]

10

37

US/Air [4]

17

42

450

Note:

3. http://en.wikipedia.org/wiki/List_of_rail_accidents_(2000%E2%80%932009) and http://en.wikipedia.org/wiki/List_of_rail_accidents_(2010%E2%80%932019)

4. http://www.ntsb.gov/aviationquery/index.aspx.  Select 1/1/2002 to 12/31/2011, County: “United States”, Injury Severity: Fatal, Operation: “Part 121: Air Carrier” & “Part 135: Air Taxi & Commuter”.

Deaths and Probabilities of Deaths

Deaths per Trillion Passenger-KM Prob. Of death
China Railways (as 1)

23.51

1.00

Japanese Railways

45.56

1.94

French Railways

21.43

0.91

German Railways

50.15

2.13

US/Air

51.90

2.21

US/Highways [5]

4,625.00

196.76

Note:

5. http://www-fars.nhtsa.dot.gov/Main/index.aspx.  In 2010, the fatalities per 100 million vehicle miles travels were at all-time low 1.11.  Use this number a reference and assume 1.5 passengers per vehicle (high-end estimate of passenger vehicle utilization).  This understates the real fatality rate of American Highways by quite a bit.

 

 

Categories: Analysis, General Tags: , ,
  1. January 7th, 2013 at 21:53 | #1

    Thanks for this great data-driven analysis, jxie. Btw, I embedded a “more” tag at the end of your first section.

  2. January 8th, 2013 at 21:33 | #2

    @jxie

    I like the simple & highly readable style of your post. For the sake of impact, may I suggest that you edit this post & put the conclusion of the study near the beginning of the post, rather than the end? Let people know right away what you concluded, then tell them the details of the analysis that drove the conclusion. That way readers can get to your central thesis right away. This is just something I learned from consulting case interview practice that can be applicable to other types of writing.

    Also, based on previous discussions stemming from previous blog posts and Pedestrian Observations (1st link below), I did some of my own calculations using similar methods (but with more accurate World Bank data), which I summarized on China-Defense Forum (see 2nd link). I estimated that you’re 30 times more likely to die on a US train than on a Chinese train. I realize rail isn’t America’s primary method of long-distance travel, and its passenger railway system is in some ways relatively primitive, but nevertheless its a noteworthy observation.

    http://pedestrianobservations.wordpress.com/2011/06/02/comparative-rail-safety/#comment-8543

    http://www.china-defense.com/smf/index.php?topic=418.msg201623#msg201623

  3. January 8th, 2013 at 22:00 | #3

    @Mister Unknown

    Thanks. Made the recommended change.

  4. January 9th, 2013 at 08:15 | #4

    @jxie
    While we’re at it, I don’t mean to be nit-picky (so please don’t be offended), but the grammar could be improved here, so as to make this piece more professional.

    “You are about 2.2 times likely to die in an American airplane as in a Chinese rail car.” – I think the grammatically correct structure would be “AS X as Y”; for example: “you’re twice AS likely to die in X as you’re in Y”. Another correct alternative would be “you’re about 2.2 times MORE likely to die in X THAN in Y”.

    The same grammar applies to conclusion #4 as well. The GMAT has seared this stuff into my brain. 🙂

  5. January 9th, 2013 at 10:33 | #5

    @Mister Unknown

    You are a champ. Just give me straight… I am a big boy — I can take it. I revised the wording of the whole post a bit.

    To be honest with you, the data portion took considerably more time for me to get it right and I was running out of my own allocated time, so didn’t do the needed proofreading. Also “2.2 times MORE likely to die”, is “3.2 times as likely to die”.

    Modern humans tend to have very short attention spans, and I purposely tried to keep the post reasonably short and to the point, so casual readers can read it fast. A lot of complexity is hidden from the post. If this post should be commented on a bit more by those who love to dig into more, I can reveal more. For instance, there is a lot more fatalities in the American Generate Aviation, e.g. in 2010 there were 450 fatalities in Generation Aviation but only 17 in the data I computed. The reason is that among the GA fatalities, there were crop duster airplanes, hobbyist airplanes, etc. I only limited to revenue-generated aviation, which I believe it’s a fair comparison to others’ passenger railways. By the same logic, I don’t think we should include fatalities involved American freight trains and metro rails, or at railway crossings. It’s very hard to get that right… but I doubt American railways comparatively speaking, is that unsafe (30 times riskier).

    The purpose of this post first and foremost is to get it right, so PLEASE critique away!

    Personally the main reason to write this post to begin with, is the annoyance. After the Wenzhou accident, an American blogger who shall remain nameless, wrote in this forum that he felt obligated to criticize because he and his family rode on Chinese rails. Also whenever you see the topic being brought up in the Internet, reporters will always bring up the Wenzhou crash in a snarky way that implies the Chinese railways are horribly unsafe, and commentators will go as fast as claim, “you wouldn’t pay me enough to ride their trains.” Well —

    I want to scream in their ears, “sons, Chinese rails are the safest you’ve ever ridden on, if you should be so eff’ing lucky.”

  6. January 9th, 2013 at 10:55 | #6

    Some observations on the data set…

    If the Japanese don’t have another major disaster between now and 2017, once past 2016, the 112 deaths will slide out of the 10-year window, and based on the methodology of this post, they will become likely the safest. The cause of the 2005 Amagasaki rail accident in essence was due to the desire to be punctual. On that day, the rail was delayed a bit prior to the accident, and the rail operator wanted to catch up the lost time by speeding up, which eventually caused the train to derail at a curve.

    If you look at the Chinese data, in the 6 years between 2002 and 2007, there were only 8 fatalities, which given the Chinese rail traffic volume, it’s very safe. From that you can’t help but draw the conclusion that the Chinese railway operations can be extremely safe. The next 4 years saw a lot more fatalities, based on the high Chinese railway safety standard established previously. A part of the reasons of course was due to the newer and faster lines and trains being introduced, there are kinks in the system that need to be worked out, and railroad operations need to be brought up to the high safety standard.

    This is not unlike the latest Boeing Dreamliner 787 problems — it’s a part of the human process to get better. Sure at this moment 787 isn’t as safe as say the proven 747, but:

    1. Make no mistake, Boeing is a top-notched plane maker.
    2. Dreamliner is such a revolutionary plane that once all the kinks are worked out, it’ll make people forget about all the old planes.

  7. January 10th, 2013 at 04:35 | #7

    @jxie
    So the railway fatality figures for China (& the others that were mentioned) – did you also have to exclude deaths, if any, that were related to non-passenger rail such as cargo, or did the data already exclude them for you? Also, the passenger-KM data, I assume (via literal interpretation of the term ‘PASSENGER-km’) that the distances traveled by cargo was not included in the KM figures, right?

  8. January 10th, 2013 at 12:04 | #8

    The source of the railway fatalities is the Wiki pages. I did some more digging and for sure it didn’t include fatalities in some freight train accidents, railway crossing accidents, and metro rail accidents in China. The passenger train accidents though, based on my extensive search, seem to cover all accidents in China. To maintain that consistency, I only count passenger rail fatalities for all countries.

    In the US, there are extensive data and reporting done by NTSB. There is no equivalent of NTSB in China, which should be a good idea to have. You want an outside agency to check the safety standards and records of MoC and MoR alike. Anyway, I doubt you have easily available data covering all rail-related fatalities in China.

    The source of WB’s “passenger-km” data for China is likely MoR’s own data, e.g. http://www.china-mor.gov.cn/zwzc/tjxx/zyzb/201012/t20101227_571.html. WB’s data matches all MoR data up to 2009. In 2010, WB’s passenger-km number is 791 million (passenger-km), but MoR’s is 876 million. My guess is that WB’s data was a projected number, and the projection was low because it didn’t factor in added passenger traffic with the new HSRs. However, I stuck with WB’s data for consistency sake.

Time limit is exhausted. Please reload the CAPTCHA.