Thursday, June 23, 2005

Are rails really safer than roads?

Tuesday's terrible train accident on the Tel Aviv-Beersheva line, with at least eight fatalities, has turned the spotlight briefly onto the subject of rail safety in Israel in general.

Passenger rail use is booming in Israel in recent years, with new routes and stations opened yearly. One of the arguments road safety advocates routinely make in favor of trains is that rail travel is much safer than the roads. I'd like to put that proposition to the test.

Conveniently, Sunday marked the launch of the new website of Israel's Central Bureau of Statistics. Much easier to use than their old site, it's the place to go for official statistics about all aspects of life in Israel. Heaven for statistics freaks like me.

One must be cautious when comparing stats between road and rail accidents, since the data is recorded in different terms. Also, most rail accidents also involve road vehicles, so there may be some double counting. Bear that in mind while absorbing the following.

As I see it, the most significant statistic regarding transportation accidents is the number of fatal accidents per passenger kilometer. It answers the ultimate question: If you travel a certain distance, how likely are you to be in a fatal accident (God forbid)?

(Alternately, one could look at the number of actual fatalities per passenger kilometer, but that statistic is more volatile due to occasional high-fatality accidents. Either way, the overall analysis doesn't change much.)

Just the stats, please
Let's start with the roads. Road vehicles travelled an estimated 38944 million kilometers in 2003 (PDF), the last year for which complete data is available. An average year recently has seen about 450 fatal road accidents. Making an assumption favorable to trains, let's assume that the average vehicle contains one passenger (it's obviously significantly higher than that due to families, vans, buses). That gives us an average of 11.6 fatal accidents per billion passenger kilometers.

Trains, meanwhile, carried an estimated 1278 million passenger kilometers (PDF) in 2003, and recent years have seen an average of 16 fatal train accidents (see PDF charts here, here and here). The average: 12.5 fatal accidents per billion passenger kilometers. Not that different from the roads, and certainly no better.

Since the average road vehicle clearly carries more than one passenger, the roads actually fare even better than that compared to trains. It also seems likely that a typical fatal train accident has more fatalities than a typical fatal car accident, but I have no statistics on that.

Admittedly, such raw statistics can be misleading. Stats for road accidents include both urban and intercity travel; no one in Israel takes the train to go down to the shops. Yet, trains do travel in and out of city centers, so they do substitute for urban transport as well as interurban.

In any case, these are the best statistics currently available (to my knowledge). And they don't support the claim that rail is safer than road. Not by a long shot.

Surprised? Isn't it obvious that trains should be safer than cars, since they travel on fixed rails, separated from pedestrians and other vehicles, without turning or changing lanes? Yet trains have no way to maneuver to avoid an obstacle. They also travel much faster than cars, and are many times heavier, generating a far more powerful collision. Nearly every train accident is fatal. They're probably rarely the fault of the train, but that's little consolation.

One of these days I'll hopefully get around to debunking the widespread belief that the roads in Israel are much more dangerous than they are in other developed countries. Stay tuned.

Update (June 26):

Over the weekend, I've been mulling this over, and I realized that some of my reasoning was incomplete. Let me clarify two points regarding the following statement:
...the most significant statistic regarding transportation accidents is the number of fatal accidents per passenger kilometer. It answers the ultimate question: If you travel a certain distance, how likely are you to be in a fatal accident?

I could have phrased this better. More precisely, comparing fatal accidents per passenger kilometer answers this question: If the same number of people travel the same distance by rail and by road, in which scenario will there be more fatal accidents?

The misleading phrase is "how likely are you to be in a fatal accident". The issue is that fatal accidents include those fatal to anyone, passengers and pedestrians alike. "You" the passenger may not be the one killed, or even the one most likely to be killed. Except for severe train accidents like last week's, fatal accidents involving trains usually kill non-passengers, such as the passengers of vehicles hit by the train.

The passengers on a train are generally much safer than passengers in cars. But, for the same reason, the train is far more dangerous to non-passengers than a car is. This is similar to one of the arguments made by the anti-SUV brigades: SUVs may be safer for their occupants, but they are more dangerous to those in other vehicles and on foot.

When you see claims that trains are much safer than cars, check whether the statistics offered for train accidents include fatalities to non-passengers. No one would publish road accident statistics which exclude fatalities to pedestrians. Why should trains be held to a different standard?

A second, related point is that the statistics on the annual number of rail accidents in Israel include both passenger trains and freight trains. I haven't found this broken down between passenger and freight, nor have I found stats on the number of passenger trains versus freight trains on the rails. Clearly, accidents involving freight trains are not relevant to the question of which form of passenger transport is safer. If I could find better data, I would use it. I expect the overall picture remains unchanged, though.


Cosmic X said...

I also was pondering this question when I heard about the accident. Yishar Koach for the analysis.

Zman Biur said...


Thanks. Please see the update I posted this morning.