Home > Economics - General > The costs and benefits of speeding

The costs and benefits of speeding

October 17th, 2003

Various commentators (notably including Tom Nankivell) have raised the issue of the costs and benefits of current speed limits and of more or less rigorous enforcement of those limits, and have asked that I consider costs as well as benefits of enforcing speed limits. I’m going to give some back-of-the-envelope numbers in an effort to inform the debate. The total economic cost of road crashes in Australia is at least $20 billion per year (This BTRE study gives direct costs of $15 billion for 1996, but an analysis based on economic welfare theory would include substantial costs (for example, in the BTRE analysis, there is zero cost in the case when a retired person is killed instantly in a crash).

The main cost of speeding restrictions is that of additional travel time. Assuming that complete abolition of speed limits would reduce average travel time by 20 per cent (very generous, since a lot of travel is on congested roads where the speed limit isn’t a constraint), and taking the BTRE estimate that value of reduced travel time is $10/hour, the total travel time cost of speed limits is around $5 billion per year.

It’s pretty obvious that abolishing speed limits altogether would raise road death rates by more than 25 per cent (the NT is the only jurisdiction in Australia without speed limits and its death rates are several times higher than in the rest of the country, though there are other factors as well as speed in play). The same argument applies, essentially pro-rata to more limited relaxation of limits. It gets harder to justify further reductions as the limit gets lower, but I’d say there’s a pretty good case for more general application of 50k limits in suburban areas to cover everything except main roads.

Turning to other costs and benefits, the survey evidence I’ve seen suggests that the disutility imposed by speeders on other drivers (particularly through tailgating, forcing their way back into traffic etc) is at least as large as any psychic benefits gained by the speeders.

Finally, on the enforcement issue, one point that has secured a lot of agreement is that variance in speed is as much of a problem as speed itself. On the assumption that there are a large number of law-abiding drivers travelling at or near the legal limit, and vehicles such as trucks that cannot safely exceed the limit, this provides a strong argument for both rigorous enforcement and tight tolerances. It also implies that excessively slow driving should be penalised.

Categories: Economics - General Tags:
  1. Tom Nankivell
    October 17th, 2003 at 22:43 | #1


    While I appreciate John’s taking the time to address the issue of the costs of speed restrictions as well as their benefits, an aggregate cost-benefit analysis of enforcing speed limits is no more useful than, say, an aggregate CBA of enforcing the entire food standards code (or the entire corporations law, the entire OHS regulatory regime or all product safety standards etc). Clearly, there are some road/food/corporations/OHS/ product safety provisions that should be fully enforced, others that should be partially enforced (or enforced in some circumstances but not others) and others that probably should not be enforced at all. But in an aggregate CBA analysis of enforcing all such provisions, the net benefits of enforcing the ‘good’ provisions may more than cancel out the net costs of enforcing the ‘bad’ or ‘partly bad’ provisions, thus giving a net benefit overall. Would this mean that we should continue to enforce the bad provisions as well as the good ones? Clearly not. So an aggregate CBA of the sort John sought to conduct, while of academic interest, does not tell us much of value about policy.

    Rather, what is required is a more disaggregated analysis of the benefits and costs of particular laws and alternative enforcement strategies. It may that, for example, that current urban speed limits should be reduced and/or enforced stringently, but that speed limits on freeways should be increased and/or enforced less stringently. Likewise, it may be that a more prescriptive regime is appropriate for heavily congested areas whilst a more principles- or performance-based regulatory regime — which provides more scope for individual driver discretion — is more appropriate in lightly trafficked areas. And it may that more discretion should be introduced into the enforcement of all speed rules to take into account, for example, differing levels of driver competency.

    Indeed, I suspect that were John to do some disaggregated CBAs of various current speed restrictions, he would easily be able to convince himself that many generate a net cost. Three examples follow.

    An obvious candidate is the 110kmh limit on the Hume Freeway which, compared to a limit of say 130kmh, can add a couple of hours of boredom to a Sydney-to-Melbourne trip with little improvement, and possibly a reduction, in safety. (A reduction in safety may arise because, to simplify, greater aggregate time on the road, and the boredom associated with operating a vehicle at around 110 on such a road, may raise the risk of fatigue/inattention-based accidents and this may more than exceed the safety risk of accidents from loss of vehicle control on such a road when traveling at around 130kmh).

    A potential alternative to (or augmentation of) the above policy option is that there is a strong case for less stringent enforcement regimes on the Hume and other non-urban freeways. Such freeways are rarely congested and the traffic is all going the same way. There are no-cross roads and no on-coming traffic to consider. Thus, a given speed variance poses less safety risk than it would on other roads where, for example, drivers need to judge the speed of oncoming vehicles when judging whether or not they have time to overtake. Thus, perhaps detection rates and/or fines should be reduced, or fine increments expanded, for such roads. (I recognise that there are issues of regulatory simplicity to also consider here, although reducing detection rates should not be difficult to achieve).

    As further food for thought, John might also consider the merits of the demerit points system. Presumably, other factors such as weather conditions and traffic density aside, the social costs of driving at a particular speed in excess of the speed limit are the same whether someone is doing it for the first, the second, the third or the thirtieth time. (In fact, someone doing it for the thirtieth time may be more experienced at doing so, which might mean the social costs are lower as they are less likely to crash at that speed). On one view, economic efficiency requires that the tax (the fine multiplied by the probability of detection) be set at the social costs (how ever high they are judged to be) and then drivers should be allowed to determine whether their private benefits exceed the private plus social costs. However, a demerit points system allows them to do this for a while, but then takes their license away! In other words, the tax on the behaviour (driving at a particular speed) increases massively at some point, even though the social costs of the behaviour are stable or decreasing. Of course, this approach might be supported if the goal were to encourage complete, or close to-complete, compliance with the speed limit. (Against, I recognize that there are also broader issues to do with regulatory enforcement resources, evidentiary requirements, encourage of pro-social norms etc that come into play here). However, as I pointed out on a separate thread on this blog, there will be cases, possibly many, where it is not only privately optimal but also socially optimal for individuals to exceed posted speed limits.

    Tom N. – cognito ergo zoom!

  2. Here we go again…
    October 18th, 2003 at 12:12 | #2

    Tom: oh my God, a pro-speed argument that doesn’t boil down to “I need my penis substitute and I don’t like government and taxation”… about bloody time!

    Interesting stuff, but there’s a logical fallacy in your para about demerit points. I’ll let you find it for yourself.

  3. Tom N.
    October 18th, 2003 at 14:55 | #3

    Tom N, 18 Oct 2003

    Dear “Hear we go again”, thanks for the compliment although I would claim that all I am doing is applying standard economic and regulatory analysis to the road safety issue. For a more comprehensive exposition of these approaches, see the ORR document I cited on an earlier thread.*

    Regarding the logical fallacy you claim exists in the last para of my earlier post, I’m afraid I can’t spot it. Perhaps I’ve confused things by: early in the para using the term “social costs and benefits” as a distinct and non-overlapping category to “private costs and benefits”; whilst later in the para using the term “social optimality” to refer to a state in which net private-plus-social benefits are maximised, thus implying that private costs and benefits are a subset of social costs and benefits.

    This raises an interesting aside about the definition of private and social costs (& benefits). From my reading of the economics literature, the terms are not used consistently. Sometimes the term “social costs” is used in effect to mean “total costs”: ie private plus non-private costs. On other occasions, the term social costs is used to describe only non-private costs. A further issue is what precisely is meant by private and non-private costs. Some economists seem to use the term private costs to mean costs borne by the agents making the decision. However, in its treatment of smoking^ and gambling#, the Productivity Commission, following Markandya and Pearce (1989), includes only those privately borne costs THAT THE AGENT RATIONALLY TAKES INTO ACCOUNT in her decision as private costs. This leaves social costs as being the aggregate of privately borne costs not rationally accounted for plus external costs (see PC 1999, pp. 4-2 if interested).

    * ORR 1994, ‘Compliance with the Road Transport Law’, Office of Regulaton Review Submission to the National Road Transport Commission, December.

    ^ IC 1994, ‘The Tobacco Growing and Manufacturing Industries’, Inquiry Report 39, Appendix B2, June.

    # PC 1999, ‘Australia’s Gambling Industries’, Inquiry Report 10, chapter 4, Dec.

    Markandya, A. & Pearce, D. 1989, ‘The social costs of tobacco smoking’, British Journal of Addiction, vol. 84, pp. 1139-1150.

  4. John
    October 18th, 2003 at 15:23 | #4

    Tom, thanks for the link to the ORR paper. While it’s cast at a level of generality where it’s difficult to disagree, I think the general tenor of the paper is misleading. The numbers I’ve put up provide a pretty clear indication that, on average a tightening of regulation (both safety standards and enforcement of those standards) would be beneficial. It’s true as ORR says, that there exists some level of safety that would be too great, but this is irrelevant if we are nowhere near that point.

    The paper also misses the crucial point that, even if the standard is suboptimal, the costs of variance in speed and driving behavior mean that it’s normally better to enforce compliance with a suboptimal standard than to allow noncompliance.

    Of course, it’s always possible to point to counterexamples to general claims about policy – for example, the point about variance doesn’t apply when there is only one car on the road. But the fact that, as you say, there may exist cases when speeding is optimal doesn’t invalidate my claim that on balance, tighter enforcement is desirable.

    To sharpen this up, notice that all of the points you (and ORR) have made apply equally well if you replace “speeding” with any other activity that is undesirable on average, say “theft”. There are cases when the socially optimal course of action is for one person to take another’s property without paying for it or asking permission, and 100 per cent enforcement of laws against theft is not desirable. But this doesn’t invalidate a claim, based on comparison of aggregate costs and benefits, that we should enforce the law more stringently.

  5. October 18th, 2003 at 16:35 | #5

    “There are cases when the socially optimal course of action is for one person to take another’s property without paying for it or asking permission, and 100 per cent enforcement of laws against theft is not desirable.”

    But this is building in an assumption: that where the rights of the individual conflict with the aggregate benefit to society, society should take priority – pretty much the basis of utilitarianism. It is equally possible to read the same argument the other way around, and conclude from the desirability of 100% defending property that occasionally society does not have a paramount place. And it is in fact a lot easier to start from individuals and infer aggregates than the other way around, not least from problems of definition of who are “we, the people” (apartheid was democratic, as it gave all citizens an effective vote – discuss).

    I find it interesting that just as when Keynes got angry with Pigou, so also did Marx with Stirner; it was an indication that their self consistent systems had come up against gaps: it was no longer possible to argue logically within their systems and still address their opposition, so they got angry. Any defence of social rights, to be consistent, has to be capable of withstanding criticism from full blown anarchists like Max Stirner or Benjamin Tucker (ones who cop out aren’t presenting coherent anarchist criticism).

  6. John Bennett
    October 18th, 2003 at 20:51 | #6

    John & Tom

    I’m happy for you to trade fire between Queensland and the ACT.

    I’d like to just throw in some brain storming on costs and benefits.


    shorter travel time increasing productivity
    John has estimated $5 billion per year based on a 20% reduction in travel time

    shorter travel time in Canberra in good weather
    I’m referring to the Canberra joke about hurrying home in the wet weather to minimise exposure to dangerous roads
    In good weather if the speed isn’t excessive or too slow thus increasing risk there is a reduction in the time spent on the roads so less risk exposure (puts a smile on my face too but when you think about it…)

    Shorter travel times reducing the incidence of driver fatigue
    I’m referring to the long interstate drives where an extra 20% in speed has people 20k closer to their destination every hour.

    Less cost to the community in speeding fines
    self explanatory

    Increased safety enforcement resources on other crash causes and breaches of road rules (eg. tailgating, RBT, and failing to keep left)
    This could erode the 25% estimate of removing speed limits but John seems pretty firm on it and it would reduce the prorata amount for less stringent enforcement. I note that the failing to keep left enforcement might address the congestion to some extent and make the 20% time saving estimate look less generous.

    A shift in culture to an increased focus on driver training that I believe would improve safety
    Clearly this isn’t the case in the NT but while NT while until a few years ago didn’t have any open road limits they do have limits on all other roads and have all the usual enforcement tools including speed cameras so don’t assume there isn’t a speed kills culture at play.

    Working on the less rigorous enforcement basis there would be more enforcement resources to apprehend more extreme variances in speeding thus creating another safety trade off.


    increased crashes
    Again I’m just going with the flow here and avoiding any extreme variance with the current theorists but I’m not personally convinced that just loosening enforcement a few k would have this effect. John has an argument in terms of variance that makes sense.

    disutility imposed by speeders on other drivers (particularly through tailgating, forcing their way back into traffic etc)
    taking it that far again just going with the flow though my view on this was hinted at above
    However I would anticipate that there would be a disutility on bad drivers when they feel pushed along

    Quantification I see as a challenge. There is a current $15 to $20 billion cost in road accidents that may be impacted upon and an estimated $5 billion in savings from shorter travel times prior to consideration of the factors I’ve thrown in.

    John is confident that there would be a greater than $5 billion increase based on a greater than 25% crashes without limits and some prorata effect for looser enforcement.

    As I said I’m currently staying out of the debate but can John can you clarify whether or not you estimate that a loosening of enforcement prorata calculation would match the time saving $5 billion?

    That’s my first brain storm. I just thought there could be a broadening of the relevant factors that are to be put in the equation.

  7. Tom N.
    October 18th, 2003 at 21:24 | #7


    While I thank John for taking the time to respond, I submit that the points he makes do not provide justification for his stance that there should be a general tightening of speed regulation and/or enforcement.

    Firstly, on the ORR paper, John says that it misses a “crucial point” about speed variances. I should point out that the paper recognises in three places that speed relativities will affect the social costs of driving at particular speeds. In terms of optimal speeds, the fact that greater variance adds to safety risk means that the socially optimum variance is lower than it would otherwise be, but it does not mean that it is zero. Thus, the paper’s conclusion that not only privately but also socially optimal speeds will vary from driver to driver remains valid.

    Even so, John states that “the costs of variance in speed and driving behaviour mean that its normally better to enforce compliance with a sub-optimal standard than to allow non-compliance”. The key term to note in this is “normally”. I agree that John’s assertion would probably be correct in relation to congested urban driving conditions. However, I doubt that it would be the case in most lightly-trafficked non-urban driving, and am even more sure that it is wrong in the case of non-urban freeways – for the reasons indicated in my earlier post.

    This highlights the broader point about how one should assess policy – at the aggregate level or at the disaggregated level. John appears to be comfortable making generalised propositions for policy on the basis of his aggregate benefit/cost analysis. He also states that my argument that there are many cases in which speeding is optimal does not invalidate his claim that, on balance, tighter enforcement is desirable. It may not, but what my arguments certainly do do, I submit, is highlight that making such claims will not necessarily be very helpful if the goal for devising good policy.

    To further explain why, let me retell a story about an old lawyer giving a speech at his retirment. Looking back over his career he remarked that, as a young lawyer, he lost a lot of cases that, given the facts of the matter, he really should have won; but, as an old lawyer, he won a lot of cases that, given the facts of the matter, he really should have lost; so, on balance, justice was done! 🙂

    John seems happy with policies that, “on average”, yield net benefits. I am too IF it is not feasible to disaggregate a policy in such a way that means it applies in cases where it provides net benefits but is not applied in cases where it yields net costs. But if it is possible to make such a disaggregation, it would be welfare-reducing not to. In the case of speed, there is no reason why speed regulatory and enforcement regimes cannot be far more disaggregated and tailored than they are at present – my first post on this thread gives a number of examples of how. For this reason, I maintain that John’s aggregate cost-benefit analysis, and the conclusions he draws from it, are of limited value when it comes to policy-making in relation to speed.

  8. Antoni Jaume
    October 19th, 2003 at 08:01 | #8

    Fuel consumption rise with the square of the speed. A raise of 20% in speed means a greater that 40% raise in fuel consumption.

    “However I would anticipate that there would be a disutility on bad drivers when they feel pushed along.”

    Whoever has written that is a bad driver, or rather an unfit to drive individual. A good driver do not expect other drivers to be good, so if you have to tailgate or, worse, cut the security distance between two car when advancing, then you are provably a bad driver.


  9. Tom N.
    October 19th, 2003 at 14:23 | #9

    Tom Nankivell, 19 Oct 2003

    Although in earlier posts I have argued that John Quiggin’s aggregate CBA on enforcing speed limits reveals little of use for policy makers, I also intend to chip away at some of the assumptions and figures that underpin the result that John arrived at.

    To start with, John said that his assumption about travel time reductions (of 20 percent) from relaxing speed restrictions is “very generous, since a lot of travel is on congested roads where the speed limit isn’t a constraint”. But if that is so, relaxing the speed limit on such roads would presumably have little or no impact on the number or severity of accidents, so John’s estimates of the costs of accidents should likewise be reduced.

    In any case, I have my doubts about his notion that abolishing all speed limits — even if that were what were proposed^ — would necessarily increase road death rates by (more than) 25 percent and that there would be a pro-rata increase in deaths as speed limits were relaxed. I know that John is just using a ball-park figure and I do not have evidence or data to back up my doubts to hand, but when I researched road safety issues in the mid-1990s I found, amongst other things, that the official attribution of accidents to speed had shaky underpinnings and, of particular relevance for long-distance country driving, that the relationship between speed and inattention/fatigue, which is inverse over certain speed ranges, was not given much official recognition. That said, I have no doubt that relaxing some speed limits would lead to an increase in deaths. However, I also suspect that in other cases (most obviously non-urban freeways) there would be little change or even a reduction in deaths and injuries up to some level of relaxation.

    Next, John states “the survey evidence that I’ve seen suggests that the disutility imposed by speeders on other drivers (particularly through tailgating, forcing their way back into traffic etc) is at least as large as any psychic benefits gained by the speeders.”

    The first response to this is that John has mixed-up the phenomenon of speeding (driving faster than the posted limit) with other phenomena (tailgating, forcing etc). Most tailgating and forcing I experience occurs in city traffic is done at under or around the speed limit. It also sometimes occurs on country roads, but speeding is neither a necessary nor sufficient condition for tailgating or forcing and, while speeders may be more likely to engage in such behaviour than others, such behaviour is primarily a rationale for encouraging compliance with courteous and/or anti-dangerous driving laws — it bears at best only indirectly on the optimal speed limit issue.

    The second response is that I doubt that John has seen any meaningful survey evidence that allows him to know how significant are the psychic benefits enjoyed by people who speed, either in absolute terms or relative to any purported annoyance-related disutility that others experience as a result of speeders speeding. Personally, my own back-of-the-envelope calculations suggest that, as reflected in my chosen speeds in different environments and the frequency with which I expect to receive fines for speeding etc, I am willing to pay at least $10 per hour to drive at 30kmh over the limit on a decent country road. This is more than triple the value of time savings reported in the BTRE study — which John asserted were the main cost of speeding restrictions. Of course, my preferences may not be representative*, but the key point I would make is not that the psychic income from speeding is necessarily large in aggregate or that all people receive it, but that for some people it is significant. (Ideally of course, regulatory/enforcement regimes should recognise this and not cause people to forgo such utility needlessly or without there being sufficiently offsetting benefits. One-size-fits-all regulatory regimes enforced bluntly will tend to find it difficult to pass this test.)

    There is at least one further aspect to the costs of speed restrictions that John has not noted. As well as lower speeds causing the trips that are undertaken to be longer and less enjoyable, at the margin some people who would undertake trips will be put off from doing so altogether. Dare I say it but this may even diminish social capital, particularly for people in rural and regional Australia! Again I do not have any formal study of the phenomenon available, but I will set out some personal anecdotes and examples in a footnote (#). I do not want to exaggerate the magnitude of these costs in an aggregate sense, but nor should they be ignored.

    In summary, I suspect (though cannot prove) that John overstates the benefits and underestimates the costs of speed restrictions. The more important point to reiterate, however, is that I do not see aggregate CBAs as being of much use for policy formulation in relation to speed. Rather, we need to examine the merits of changes to policy at a more disaggregated level.

    Tom N. – cognito ergo zoom



    ^ In my view, serious reform in this area would need to be selective (mainly for country and freeway driving), phased in, and possibly accompanied by other measures including education/information measures and greater enforcement of other rules, such as lane discipline and anti-dangerous driving laws.

    * There are several reasons why my willingness-to-pay for speed may not be representative. I have above-average discretionary income and, while not a V8 driving or Bathurst-following rev-head, show more interest in the recreational/therapeutic value of (some forms of) driving than most of my peers. Further, although originally from the country, at present I drive on country roads only around once a month on average, mainly between Canberra and Sydney or Canberra and Wagga (or just a quick fang through the Brindabellas). Presumably, were I a more frequent country traveler, diminishing marginal utility would kick in and I would be willing to pay less than $10 per hour for the pleasure of driving at a reasonable pace. Alternatively, were I a city-bred person who had ventured out of the big smoke by car very infrequently, or someone who placed great store on obeying the law for its own sake, or just someone who was strongly influenced by the “speed kills” messages promulgated by the RTA, I might in fact get psychic disutility from driving at speed and thus may not in fact be willing to pay anything to exceed the speed limit.

    # For people in country Australia, speed restrictions can have costs in the form of trips not undertaken, and family and social interaction not had. To provide some personal examples, several years ago I curtailed the number of weekend trips home from Canberra to Parkes to visit friends and family during a period when I was close to my demerit points limit, because sticking to the speed limit turned a highly enjoyable 140 minute trip into an extremely boring 200 minute trip – two of which had to be squeezed into a weekend. Similarly, these days I would get down to Melbourne much more frequently to visit my friends were I able to do so at a reasonable speed (say 140-160kmh) on the Hume freeway. As a final example, from questioning fellow university students in Armidale about this issue, I found several who (in some cases due to having a pathetically slow car, but the point is the same) limited their trips home to Sydney or other destinations due to the time that would be entailed, as did I.

    PS: In his post Antoni Jaume notes that traveling at higher speeds raises fuel consumption. This is of course correct, but fuel consumption is a private cost and thus per se is not relevant for public policy. Of course, it may be relevant if the environmental externalities associated with burning fuel are not reflected in the price (which already incorporates a significant tax component), but were this the case the first best policy solution would of course be to increase the tax, not to limit speed.

  10. John
    October 19th, 2003 at 14:34 | #10

    Fuel costs are private, and therefore do not form part of the externality analysis that justifies speed limits. Nevertheless, they are relevant in assessing likely private willingness to pay for more relaxed speed limits.

  11. John
    October 19th, 2003 at 15:46 | #11

    Tom, at 160 km/h braking distance is over 100 metres for top cars in perfect condition, 150-200 metres for many standard cars, more if brakes aren’t working perfectly. Reaction distance (to interpret possible dangers over 100 metres away) is over 40 metres per second.

    This might be sustainable on the highest quality sections of a German autobahn – three or four lanes each way, long acceleration and deceleration lanes for on and off-ramps etc, but there is nothing comparable to this in Australia. (Contrary to myth, the majority of the autobahn is speed-limited).

    Given the large number of trucks using major Australian highways, even a four-lane highway effectively only has one lane open to traffic above 100 km/h.

    It’s worth recalling that, contrary to the assumption implicit in the ORR article that traffic authorities are mindlessly reducing speed limits, the 110 limit on freeways is an increase on the 100 km/60 mph limit that applied in the past. I think you should explain why you think they got it wrong in not setting a higher limit. The US evidence (little enforcement of speed limits on a road system better than ours, but not autobahn-grade) suggests that there is a substantial cost in additional lives lost.

  12. Tom N.
    October 19th, 2003 at 20:10 | #12

    Tom Nankivell, 19 Oct 2003

    If you drive along the Hume Freeway, you will find that at many times you have an essentially autobahn standard road virtually to yourself or, at least, with only a few vehicles within sight at any one time travelling in your direction. Often you also have unrestricted views of the road a kilometre or more ahead. If under a relaxed speeding regime most cars were travelling at, say, 130-140kmh, then doing a speed of up to 160kmh^ would involve a closing velocity of at most 30kmh. Even coming up to vehicles doing 110kmh would entail a closing velocity of only 50kmh. The reaction and breaking distances that John mentioned are, I suggest, more than sufficient for reasonable moment-specific safety in such conditions. Further, the reduction in travel time and boredom on long trips would bring its own safety dividend.

    (That said, I recognise that there are some non-urban freeways – the Princes freeway between Sydney and Wollongong, for example – that are far more congested and for which a lower speed limit, AND tighter enforcement to reduce variances, may well be appropriate – at least during peak times of the day.)

    I will also assert that there are many stretches of country roads – west of the Great Dividing Range and particularly West of the Newel – where it is also sufficiently safe to cruise at up to 160kmh in terms of one’s own vehicle control, depending of course on the vehicle and driver competency etc. However, there is obviously more chance of “unexpected events” – such as Kangaroos, entering vehicles etc – arising in such conditions, plus the issue of the occasional on-coming vehicle, and as I mentioned before variances are more important on these roads (holding the no. of vehicles constant), so the socially optimum speed would presumably be lower for these roads.

    Of course, I recognise that 160kmh may well sound frightenningly fast to a populus that has been dumbed down to accept that anything over 100-110kmh is dangerous. As mentioned in my previous post, I would not recommend increasing speed limits to allow such speeds overnight (and would want to study the evidence closer first before settling on a final maximum speed limit^). But whatever the appropriate maximum speed on such roads, I am confident that a proper analysis would find that they are well in excess of 110kmh for most non-urban freeways, and 100kmh for many country roads. I also beleive that a sophisticated, disaggregated analysis would find that the current control of variances is too tight on some roads such as non-urban freeways and certain lightly-trafficked country roads.



    ^ I should point out that, by stating in the footnote to my earlier post that a reasonable speed on the Hume might be 140-160kmh, I was not implying that the limit should be set at 160kmh. As I have pointed out previously, there are externalities associated with speed, and a regulatory regime should take these into account. One way would be to set a limit at, say, 130 or 140kmh on such freeways and then levy fines for breaches over the limit. This would still allow one to do 160 where appropriate, as long as one’s willingness-to-pay was greater than the fine/detection regime charged.

  13. John Bennett
    October 19th, 2003 at 20:49 | #13


    There’s usually not a lot of reason to stop on those roads and usually a very large distance to see the things you need to stop for. My understanding is that they usually have design speeds of 130 and allow for a reaction time of 3 seconds. How does that fit in with the maths you have done?

    I have heard comments about the finely built sections of the German autobahn before. Currently I believe it is a rumour. Can you assist? I have not driven in Germany. The basis of my present opinion is two fold:

    1. I have seen photos of the autobahn that made it look less than impressive. I’ve seen better roads in Australia than I’ve seen in the photos and authorities don’t have to remove snow and black ice from our highways and motorists don’t have to encounter it.

    2. I’ve talked to Europeans who indicated that our better roads are better than any they’ve seen anywhere in Europe. (If you drive to the Gold Coast from Brisbane you encounter three or four lanes each way on some sections of the Pacific Motorway.)

    I’m open to you correcting my belief as I am not absolutely certain and don’t want to entertain false beliefs. But to do so you would have to advance something more compelling than points one and two. I presently don’t know the basis of your belief.

    You commented that contrary to popular belief most of the German autobahn is speed limited. Can I clarify if speed limited is the same as having a speed limit in the way that we do. I’ve heard that they intermittently impose temporary speed limits on the autobahn (eg. if faster traffic is approaching slower traffic from behind).

    “Given the large number of trucks using major Australian highways, even a four-lane highway effectively only has one lane open to traffic above 100 km/h.”

    Given the lack of enforcement of anything but speeding you are lucky to have any lanes open to anything above 100km/h. Sometimes you can achieve it in the left lane because the road hogs prefer to block off the rest. This lack of lane discipline results in other illegal behaviours such as the unsafe overtaking I alluded to and tailgating.

    I note that when 110kph zones were introduced in Queensland Bredhauer announced that a 7 year trial on over 2000km of road showed that the trial increase from 100 to 110 reduced accidents.

    I further note that the Queensland parliamentary road safety committee in 1991 seemed to suggest that most people were driving closer to 110 than 100 so the speed limits should be raised to more closely approximate the speed on those types of road. On a similar vein I’m sure I read a NSWelsh paper that indicated that speed limits were originally set to reduce variance. Has anyone else encountered this idea?

    “Whoever has written that is a bad driver, or rather an unfit to drive individual. A good driver do not expect other drivers to be good, so if you have to tailgate or, worse, cut the security distance between two car when advancing, then you are provably a bad driver.”

    Antoni I wrote that. I wasn’t actually thinking of tailgating as I believe that there would be an enforcement shift that would reduce road hogging and tailgating. It was more the idea that if limits were raised there would be a uniform stream of traffic moving faster than an incapable driver would feel comfortable with as I believe most people find present limits too low. This effect might make the incapable driver feel pressured to go faster. It was just a brain storm. I also believe that there would be better driver education so that would address the incapable driver issue and remove the problem.

  14. robert
    October 20th, 2003 at 16:47 | #14

    This is a breath of fresh air on the speeding issue – I agree with pretty much everything you’ve said. However, I wonder about a couple of things.

    You said that you think the speed limit on country roads and freeways should be well above current levels, but I’m wondering how high you think it should go? BTW, one counter-consideration that you have not mentioned is that the variance issue relates not only to cars but also to slower moving vehicles – tractors, cyclists and the like – that you sometimes see in country areas, on country roads and highways if not the freeways.

    Also, you seem to dismiss that there may be areas on the coast or even in urban areas where speed limits are too low, yet I could point to several coastal roads where in my judgment this is not so because they are good quality and generally do not carry much traffic relative to the road quality.

    I’m also wondering if your ideas have had much airing? You said that you looked into road safety issues in the middle 90s so I take it that you worked on the Office of Regulation report. But look at how speed limits are set and policed today and its hard to see much sophistication in the approach, so what happenned – was it shelved for political or revenue raising reasons?

  15. Here we go again…
    October 20th, 2003 at 20:20 | #15

    Tom: Nah, that wasn’t it – but thanks for clearing that up, it was a bit confusing.

    The logical problem I see re demerits is that you say that the social costs of the behaviour is stable or decreasing, but this does not actually flow from your argument. (You do say that someone who is speeding for the 30th time “may” be better at it, and I spose therefore less likely to crash, but you provide no evidence for this and it’s by no means self-evidently true.)

  16. John Bennett
    October 20th, 2003 at 21:40 | #16

    Here we go again

    It seems self evident that repeatedly exceeding the speed limit has a stable or decreasing social cost. The behaviour is the same and as you highlighted they are getting more practice at it.

    It is widely believed that learner drivers would benefit from increased exposure to driving in different situations before they are ‘let loose’ in a motorvehicle. Many consider this to be self evident.

    While the learning curve should be more level for someone exceeding the speed limit isn’t it analogous to have repeated experience with that type of driving? The drivers are controlling a car while taking an active role in determining the speed at which you encounter the environment? A driver who drives for the speed limits is practicing matching their speed with the number on the sign. A driver who drives for the conditions is practicing matching their driving speed with the surroundings. Wouldn’t the ability to drive safely for the surroundings improve if practiced?

    At the very extreme are race car drivers who emphasise speed and develop skills at managing extremely high speed without crashing. Would you advocate driving around a race track at slow speeds dictated on signs as suitable training for races? That would be silly wouldn’t it? Clearly they need to practice by choosing their own speeds as they drive so that they develop the skill to adapt their high speed to the environment by practicing.

    A driver who never races does not need to take things to the extreme because they are not preparing for a race. That does not mean that they don’t learn anything or develop any skill.

  17. Tom N.
    October 21st, 2003 at 00:27 | #17

    Tom Nankivell, 21 October

    Thanks for your comments HereWeGoAgain. I know this is not a major issue, but let me clarify the reason for my view of stable or decreasing marginal social costs (and please forgive me if I over-elaborate).

    Other things equal, the social cost (mainly the risk of an accident multiplied by the costs to other drivers hurt, the public health system etc) of driving at a particular speed above the limit (say 130kmh on a 100kmh posted road) is the same whether I do it today, tomorrow or next year, and the social cost per time that I speed does not increase (even though the total social costs of me speeding three times is triple the total social cost if I speed just once). It is in this sense that I said that the (“marginal”) social costs of speeding are likely to be stable.

    In fact though, it seems to me that social costs are likely to fall with each additional episode of speeding, because the driver becomes more competent at handling the vehicle at that speed as result of greater experience.*

    You say that the proposition that frequent speeders are less likely to crash in any one instance of speeding is “not self-evident”, so to make it less abstract, let me tell you of the example of a relative of mine. He is a home appliance sales and installation/service agent in rural NSW and, like many sales reps and country service people, he often spends several hours in a day driving on country roads from job to job. Both to maximize enjoyment/minimize boredom and to reduce traveling time (which affects how many jobs he can do in a day and thus the financial viability of his business), he drives on such roads at speeds often well in excess of the speed limit^. Because he has to spend so much time on the road, he has invested in learning advanced driving skills to sharpen his ability to handle incidents that may arise. But even if he had not done this, I would be willing to wager that his ability to handle a car at 130kmh or more in an emergency would be far better than the average driver’s. In other words, his “need for speed” made him invest in sharper driving skills, but experience itself has also given him skills that now make him safer at a particular speed, meaning the social costs each time he speeds is less than the social costs each time a less experienced driver speeds (other things constant). Because my relative speeds frequently, he also gets more speeding tickets than the average person, as you would expect^. But the penalty does not fall to reflect the lower social costs entailed, nor even remain constant (in terms of fines). Rather, it actually rises (due to demerit points).# It is interesting to consider some of the potential effects of this. If he approaches (or exceeds) his demerit points limit, he may well have to hire someone else to do his out-of-town jobs, or conscript another family member to drive him. As well as adding to costs, this will result in a less experienced driver behind the wheel. The effects of this on safety would potentially be negative.


    * That said, the relationship may not be so straightforward, because it is possible that the first couple of times driving at high speed the driver may be very alert and ‘careful’, but as the driver speeds more frequently she may become more confident and less alert. It is of course an empirical question as to which of these two effects would dominate, but my sense is that, after a while, the experience of high speed driving would more than outweigh the possible additional alertness of speeding the first couple of times.

    ^ This is not to say that he “ignores” the speed limit. Rather, the risk of penalties causes him to reduce his speed, often to much lower than he would otherwise do. However, even with the risk of penalties, his optimal speed can still be a couple of increments above the limit.

    # As mentioned in my earlier post, I recognise that there may also be broader considerations than just setting the penalty to reflect the specific social cost associated with the breach — to do with regulatory enforcement resources, evidentiary requirements to determine likely, but unobserved, behavior patterns, and the encouragement of pro-social norms — that bear on the appropriate enforcement regime.

  18. Tom N.
    October 21st, 2003 at 00:49 | #18

    Tom Nankivell, 21 October

    Thanks for your comments Robert. Regarding my formal involvement in this issue, I have worked as a public policy adviser on various matters dealing with safety risk and regulatory design and enforcement, including in relation to product safety standards, food safety, smoking and pharmaceuticals as well as road safety. I also worked in a road transport regulatory agency for a brief stint, dealing mainly with heavy vehicle speed and driving regulations. In terms of the Office of Regulation Review paper, it was only a submission to the National Road Transport Commission and, while it was “considered”, its approach did not subsequently find its way into the Commission’s policy. Without going into details as to why, suffice it to say that my sense at the time was that the road safety field was dominated by entrenched state bureaucracies and that the dynamics of policy formulation often lent itself to conservatism, ‘maximum visible regulation’ and the taking of an unduly narrow public safety-focused approach to driving regulation, rather than a broader economic-style approach to road use, safety and people’s individual and collective wellbeing.

    Robert also asked how high do I think speed limits should be. Let me preface my answer by reiterating that I would want to have a closer look at the latest available evidence before coming to a firm recommendation. I would also take a phased approach, lifting the limits by a little first and then, after a period for adjustment and all going well, by more later. And I would also want to reform other aspects of our driving laws and enforcement practices. That said, my sense is that over time non-urban freeways should see a posted limit of up to 140kmh and good quality sections of country highways and roads should see a posted limit of up to 120kmh. (I accept that there may well be a case for raising the limits on some roads in coastal and urban areas too, at least during non-peak periods^, but I agree with John Quiggin that a reduction to 50kmh for all those residential streets still at 60kmh is probably warranted).

    Robert mentioned the issue of very slow vehicles such as tractors and cyclists. Having ridden my bike through the Snowy Mountains and across the Nullarbor Plain, I am not unaware of that there is a safety risk involved in being a touring cyclist and that higher speeds on country roads could potentially increase the risks (though not necessarily — on the Nullarbor the key factor is that drivers be alert enough to see you (and vice-versa), not how much over 100 they’re doing). However, as I’ve mentioned earlier, the task is to find the speed regulation system that maximizes welfare; not that simply minimises safety risk.


    ^ Greater use of variable electronic speed limit signs to deal with traffic (and possibly weather) variations for appropriate major routes would seem a sensible idea.

    PS: Dear HereWeGoAgain and John Bennett, I did not see JB’s last post about decreasing social costs before I put mine on the thread. Sorry for the partial duplication/repetition of thoughts.

  19. John
    October 21st, 2003 at 06:55 | #19

    Tom, given the multiple threads on this, I may have missed something, but I don’t think you’ve responded to my big question on highway speed limits – what about trucks?

    I don’t believe that trucks can safely sustain speeds of 140 km hour and, in any case, most Australian highways are sufficiently hilly that they can’t physically sustain this speed. In most cases 100km is an appropriate limit

    But then the point about variance comes in. If there were no trucks on the Hume or the Newell, a limit above 110 might be OK, but in reality there are lots of trucks.

    I don’t deny that there are times and places where the limit is too low. But the best way to address this is not to try and fine-tune the limits (this engenders the kind of complaint, already seen on this board, about limits changing randomly to trap drivers) but to focus enforcement efforts on areas that have experienced high accident rates.

  20. October 21st, 2003 at 10:41 | #20

    This idea from Tom N. that a road can be seen to be safe and therefore we should be able to go faster ignores the physics. [See http://www.nhtsa.dot.gov/cars/problems/studies/brakes/Phase2/COVER.html%5D

    A week ago today, a neighbour, and incidentally a travelling salesman, was killed in a headon collision at Colac. I do not know the exact circumstances of this accident, but I do know that I only travel on the M1 from Geelong to Portland when I have to. Faster than 100 kmh seems ridiculous to me.

    And what about at night? A variable speed limit day and night, or the same for all conditions? Yeah well in black cow territory in my neck of the woods, it is my opinion that anyone who drives faster than about 100 kmh at night is a fool, no matter how clever they might be otherwise.

  21. John Bennett
    October 21st, 2003 at 20:48 | #21


    I found your link interesting but it was hardly a physics of speed article. It was more a physics of braking paper. This has limited relevance to safety unless you drive like the people in the DoT ads. I’ve discussed driving at physics at a very basic level using tennis balls to illustrate the motion.

    Here is an article on the physics of driving that attempts to consider the issues more thoroughly.


  22. October 22nd, 2003 at 05:35 | #22

    Thanks for the link John B.
    The problem with this analysis is that the most dangerous roads in Oz seem to be open country roads -think kangas thru the windscreen, or headons with a logging truck, and then the low traffic density model in the first part applies. In fact *confirms* my contention that raising the speed limit on country roads would be a mistake.
    I will look at the paper a bit more, as the old joke about halving exposure on intersections by
    doubling speed seems to apply, and is obviously nonsense.

  23. October 22nd, 2003 at 06:41 | #23

    Incidentally, John B. physics has more than one branch. The link about stopping distances and the like above is the *measurement* or empirical side of physics. Mathematical models are essential, but often enough real-world tests show up flaws in the theory. At an off-road course I did a few years ago I learned that pumping the brake quickly on gravel roads –not letting the brakes lock up– allowed a car to be stopped quicker. I did it myself. I bet this idea did not come from a theoretical model.

    Also increasing the speed limit from 100 to 140 kmh *doubles* the kinetic energy of black cows, kangaroos, motorcyclists or logs coming over the bonnet at you. [140^2/100^2]

  24. John Bennett
    October 22nd, 2003 at 11:39 | #24


    I don’t see that the analysis has problems but it is written by someone using American roads as a guide.

    Perhaps the most dangerous roads aren’t the place to be raising the limits.

    I’m not clear on how a low density traffic environment with kangaroos and logging trucks that you can head on with create a problem for higher speed although they do make the analysis less relevant.

    The traditional view of people who drive in kangaroo environments regularly is that if you drive to slowly they have an awful habit of playing chicken at the front end of your car. I can confirm my personal experience engaging in a lot of swerving at the speed limit. The popular experience is that kangaroos move fairly quickly but they usually don’t react in time to get in front of fast vehicles. I’m sure you see how this relates to your kinetic energy argument.

    As regards the logging truck a head on implies overtaking. It takes 1km at the speed limit to overtake a car going 90 and 2.1km to overtake a car going 95. Many people consider it prudent to move fairly efficiently to get away from the wrong side of the road.

    Please note that real life experience also often confirms empirical physics. ( ;

  25. John Bennett
    October 22nd, 2003 at 14:39 | #25


    I didn’t properly address your post and owe you an apology. I didn’t see the wood for the trees for a moment. You brought up the issue of kangaroos etc. and I became so enthusiastic about discussing your examples that I missed your point about the open country roads. On those roads the learned author of the paper considers that absolute speed becomes more significant.

    You are correct in that the paper differentiated between different levels of traffic density when making the calculations.

    You are correct that it doesn’t support higher speeds on low density outback roads. The primary focus of the paper is the types of roads that most people encounter regularly where speed difference from other traffic is the primary speed issue. On low density roads the learned author of the paper basically contends that going faster increases risk when there is no other traffic around.

    This in turn, as you suggest, supports your view that raising speed limits on out of the way country roads would be a mistake.

    Trading off against the physics calculations that he makes are the fatigue factor, and the single lane in each direction making you need to drive on the wrong side of the road factor that we experience on those types of roads.

  26. October 23rd, 2003 at 08:52 | #26

    Thanks for that confirmation about open roads John B. I have been thinking a bit about the “speed doesn’t kill” but drivers do meme that is so common in my neck of the woods [Portland Vic.] and I was going to try to argue logically –if i could :>( but now I think the paper you provided the link to might be a better starting point. Thanks for the discussion

Comments are closed.