Stuart’s two posts on traffic volumes in the last couple of weeks have highlighted the indisputable fact that we’re driving less, that we’ve been driving less for quite some time now and perhaps most interestingly, that us driving less is actually not a bad thing for the economy. The best data available is for state highway traffic volumes, where we can see three distinct phases of traffic growth/decline since 2008:
The data is from NZTA, and I have just overlaid the red lines and text separating out the phases. The key coloured line is the pink one, which shows the three month rolling average for all vehicles compared to the previous year. It tracks the green line, which is for light-vehicles, quite closely.
- Phase 1 is pretty much 2008, and starts with a dramatic decline in volumes in the earlier part of the year – with the rolling average for all vehicles in July bottoming out at around an 8% decline from the same months in 2007. In the later months of 2008 the decline reversed, but it took until 2009 before we saw ‘month on month’ positive growth rates re-emerge.
- Phase 2 is 2009 and is pretty much the only time when we actually saw higher traffic volumes compared to the same month the year before. However, a key reason for this is obviously because it was recovering from the massive declines in 2008. And in actual fact, many of the 2009 increases weren’t big enough to offset the 2008 declines. For example, traffic growth in June 2009 was 2.5%, but the decline in June 2008 had been 6.6%, meaning the 2009 numbers were still 4.3% below 2007 levels. Overall, only January and December 2009 had higher traffic volumes than the same months in 2007 – the rest were lower.
- Phase 3, which I think we’re still in, covers all of the last two years and is actually pretty boring really, showing pretty much no increase in volumes, although the decreases are not as dramatic as in 2008. After the huge fluctuations in 2008 and 2009 it has been a calmer period, but certainly has not reverted to the long-running trend of pretty constant growth in volumes that had occurred up until 2008.
Before I get onto the cause of these trends, I think it’s worth noting the differences between heavy vehicle trends and the general trend. Heavy vehicle volumes held up a bit better in 2008 (although the trend was downwards) than for general vehicles, but didn’t recover until much much later in 2009. In phase 3, heavy vehicle volumes have increased significantly more than general vehicles, although even they have tailed off in the last few months.
There are likely to be multiple causes for these trends. Stuart’s posts highlight changing demographics, changing cultural attitudes towards cars, technological change, transport saturation, ongoing urbanisation and rising transport prices as all contributing to this fundamental shift – which on a per capita basis has actually been quite a significant decline in the amount we’re driving. It’s hard to get good data on most of those contributing factors, except for one that I think has had perhaps the most immediate impact over the past few years: fuel prices. Using the handy AA Petrolwatch information, we can see how the price of 91 Octane fuel has fluctuated over the past four years (not adjusted for inflation by the way – something I probably should do for future posts): 
We all know the story: in the first half of 2008 oil prices spiked, sending petrol in New Zealand over $2 a litre for the first time. Then in September 2008 the global economy tanked and oil prices crashed – with petrol hitting a low of $1.33 a litre in December 2008. Prices held relatively steady (though generally increasing) throughout much of 2009 and 2010, before increasing in early 2011 (Libyan crisis?) and have stayed above $2 a litre ever since – most recently propped up by concerns over Iran and a slowly recovering global economy.
So what happens if we overlay the graphs on each other? Well, to further iron out some fluctuations in traffic volumes I have taken a 6 month rolling average, then overlaid that onto the graph above (though the horizontal lines relate to the volumes to make it clear where zero is):
What seems to correspond best, just glancing at the graph above, is the relationship between petrol prices and the slope of the traffic volumes graph. I guess that’s natural when you use rolling averages, because it takes some time for the impact to really show through.
I’m not much of a statistical whizz, but overall there does seem to be quite a clear link between the two – showing that higher fuel prices really are a significant contributor (in my opinion) to declines in traffic volumes. With the likelihood of fuel prices increasing in the future seeming higher than the likelihood of them decreasing, it seems a fairly safe assumption that the stagnation in traffic volumes isn’t going to end any time soon: terrible news for NZTA and the Ministry of Transport who are trying to justify the spending of billions on new motorways, but great news for the rest of us as chances are congestion isn’t going to increase much in the future.
Processing...
It seems the data above is showing a sudden reaction to higher prices in 08, followed by a return to old habits on the subsequent collapse of oil prices with the world economy. Then as oil prices returned to their relentless upward march we see a sustained attempt by significant numbers of people to find other ways to get around- to drive less. Outside of say suddenly losing your job and no longer driving to work, driving less is quite a hard thing to achieve in a country with as few alternatives to driving as NZ. I’m picking this as a trend to stay, because while oil [the source of 98% of our transport fuels] price will fluctuate [thanks Gerry] there is absolutely no sign of any breakthrough to the global demand-supply squeeze. Everyone is pumpin’ and frackin’ like mad and only the slow collapse of consumption in the west has prevented even more damaging prices.
And remember our high exchange rate has softened the upside of oil price so the charts above in NZ petrol price are less dramatic than the oil price charts.
What else are we seeing? Big rise in PT use and big upward track on inner suburb dwelling prices and rents. People ain’t dumb, they want to protect themselves from their vulnerability to the ever rising costs of auto-dependency.
There seems to be enough data now for NZTA to be doing “what if” scenarios for proposed projects and varying petrol prices. No excuse for not doing this.
Yes Cam, as well flat or even falling vehicle use possibilities. It would show, of course, that all the current RoNS BCRs are at the extreme upper end of any range.
In their pace and breadth the RoNS is a radical plan; Joyce is a sort of visionary, a strange backwards looking one but still one of those politicians who believes they can force their will on the world.
Man versus world: Will end in tears.
Of course there is. The Minister won’t be asking, because the Minister doesn’t want to know.
I see some interesting Question Times ahead for Julie Anne Genter and Phil Twyford as they pursue ministers about the modelling being done to account for the persistently-higher-than-projected fuel costs.
Is this Auckland data, or NZ?
Buses count as heavy vehicles right, so how much of the heavy km’s are actually due to our growing bus frequencies?
It is NZ data though I might do the same with just Auckland in a future post so I can run a comparison with PT patronage too.
Laying two graphs with different axes on the top of each other to make a comparison is not a valid method. Think of the different items you neeed to chose to try to make them look related: how the axes are aligned (for some reason you have aligned 0% with $1.65) and the scale of the two axes (2% traffic growth is equated with about $0.50c petrol price). Now try changing these things and see how they look — it’ll be completely different. And then look at a longer time series and you’ll probably decide something completely different again. Remember that we are prone to finding patterns in things. Often none exists.
What you probably want to look at is the cross-correlation between the two time-series.
And if you really think the rates of change are related, why not deal with those (take differences between consecutive months)?
If you want to post the raw data here, others could try to do something with it.
Agree David, particularly it is difficult to compare an absolute figure (petrol$) with a rate of growth (traffic volumes). To be more informative the absolute figures should be compared and petrol adjusted for inflation as mentioned.
If the absolute traffic volumes were available then I agree it would be good to use them. Sadly they are not immediately available.
The purpose of this post isn’t to statistically prove a relationship between petrol prices and traffic volumes. It is simply to show that a logical assumption (as prices go up people may drive less) has some general support when you look at the data.
Okay fair enough, it is certainly interesting to see that there is that support in the data. What has caused the significant turnaround in growth in mid2009?
2009 was the rebound from huge drops in those same months in 2008.
Going in to the CBD tomorrow. Will catch the train. Why would I want the expense and hassle of driving in and parking. Trains are a much more enjoyable transport mode. As one of many NZers who have lived abroad (london) and experienced PT over there, I am glad to be slowly getting the same options at home. I would expect many feel the same way.
As important as the ‘why’ is to ask what effects this change in travel patterns is having.
It’s impossible not to notice that this is the first fatality-free Easter weekend since records began in the mid 1950s. Wonderfully, unbelievably, the first ten days of April haven’t seen a single road death. Last year had a lower road toll than any since 1952 (Brownlee, Tolley), with 90 less fatalities than the year previous. That’s; 90 people who’re alive, hundreds of people saved serious and debilitating injury, and thousands saved from the loss of a loved one. It’s hundreds of millions saved; in accident costs, ACC costs, income earned by working NZrs, and in foregone spending on gold-plated roads.
There are of course many reasons why the road toll has decreased; safer cars, fewer young drivers, less tolerance for speed, alcohol, and illegal drugs, road safety improvements, and better driver attitudes (caused in part by policing and advertising) all contribute significantly. It would be dangerous to frame a trend in terms of a single variable. But the last time the road toll dipped so significantly was in 2008. That was also the year in which petrol spiked, and vehicle kilometres plummeted. It was also a year in which speeds decreased, in part due to fuel prices.
Not all kilometres are equal. Long-distance trips on the open road made by city drivers, and young people driving for pleasure (indulging in intoxicants or speed) are disproportionately responsible for death, much more so than rush-hour commuting at below-fatal speeds. Both are disproportionately sensitive to cost increases. Lower levels of driver licensing among young NZrs is also contributing to reduced numbers of inexperienced drivers on the road (they may however show up as inexperienced drivers at higher ages), and both these things are showing up in massive decreases in deaths of young drivers. From memory, 2011 saw 20 deaths of 15-19 year-olds, compared to 44 the previous year. Figures for 20-29 year-olds are similar.
I don’t think it would be unreasonable to run the quarterly road-crash statistics for the last few years against price and distance. If there is a relationship, you’d expect it to show up.
TL:DR – Less driving helped save 90 lives.
Absolutely agree George. It would he interesting to see some statistical analysis of the connection undertaken in a detailed way and perhaps used to estimate the impact on road safety of further increases in fuel tax.
If we are driving less then Cameron Brewer’s congestion charge on the motorways will be a lead sinker 😛
A bit off topic, but people from Howick / Botany would be driving even less into the CBD if this was built http://www.times.co.nz/front-page-feature/dreams-of-beach-wharf.html?limitstart=1 and was well serviced by ferries (with a connecting bus)
See todays Herald article from Brian Fallow about a ANZ Senior Economist has developed called “Truckometer” which purports to link the current and future economic wellbeing of the economy with the amount of light and heavy vehicle traffic on selected roads.
Seems to me that this economist may have put two different axis on the same graph and by the “both delicining at the time trend” seen on their graph, assumed a causal correlation when there isn’t one.
If in fact traffic is getting efficient for each GDP $ as has been implied in other posts here, then traffic levels dropping is a Good ThingTM not a reason for doom and gloom as the TruckOMeter indicates is the case.
I’m the author of the Truckometer. You can read the full report at anz.co.nz. Go to Commercial/Institutional then to Economic Reports. There are much longer term charts in there. With different axes.
I’m the author of the Truckometer. You can read the full report at anz.co.nz. Go to Commercial/Institutional then to Economic Reports. There are longer term charts in there. With different axes. 😉 They suggest the relationship between GDP and traffic has not changed.
Hi Sharon have seen this?: http://greaterakl.wpengine.com/2012/04/02/14-year-old-newsflash-kiwi-economy-waves-good-bye-to-state-highways/
Interesting thanks. There is some bluriness around correlation and causation in this thread. Just to be clear, my work draws absolutely no causal link from traffic to GDP. I just want to exploit the correlation to predict GDP. If it were causal I’d go drive back and forth on my best road for a few days and save the economy from recession.
Here: http://anz.co.nz/resources/9/c/9cd9b4804ad401a6a726ff415e15c706/ANZ-Truckometer-20120410.pdf
Some interesting charts there, you seem to have access to better NZTA data than the rest of us….!?
Technical questions: NZTA defines buses as heavy vehicles, as you use the word ‘trucks’, I take it you somehow separate all passenger vehicles out of your data sets? Also how did you select your included measuring sites?
I get that you are mining their data to try to find an indicator of likely future economic performance but it is also telling, is it not, that there is a very big gap now between traffic growth, light and heavy, as actually recorded, and the numbers NZTA and the MoT use when building business cases for their favoured projects? Especially the RoNS. It’s been at best flat for seven years now, and this is despite billions of dollars of new construction in State Highways which are supposed to make the use of these assets more attractive, no?
As an economist aware of the trends wouldn’t you say it’s about time they rewrote the handbook to include lower, or even negative, growth possible modeling too? Looks like we may be pouring billions of dollars away because of a refusal to see what is plainly in front of us…?
Steve Jobs said ” Things have to be believed to be seen”: Looks like a lot of delusional group think going on in the transport ministry and agency right now.
I will not comment on political issues.
To answer your technical question, the heavy traffic includes buses. I did look at some more detailed breakdowns but they weren’t much more informative.
I chose my sites through classic data mining. I imposed no restrictions based on priors about which roads “should” matter. See the technical notes in the release.
How is an apparent disconnect between modelling used in policy agencies and the real-world data produced by those agencies a “political issue”? Surely you’re seeing evidence through your examinations that has made you question the assumptions of endless traffic growth? If you haven’t been asking yourself those questions, are you really looking at all the data?
A very interesting answer! So Sharon you are saying that the numbers used by MoT and NZTA in their modeling are decided at the political level and are not objective or evidence based inputs?
Hi Sharon,
Thanks for clarification.
I think you need to talk urgently then to Brian Fallow as he is clearly believing based on his Herald piece that there is a definite causal link between traffic and present and future GDP in the Truckometer.
I don’t deny the present economic state may be measured by the TruckOmeter its the delayed “causation” effect Brian Fallow talks about that seems wishful.
In the same way that metrics for EFTPOS and Credit Card spending data can be used to show the current (or recent past) state of the economy, but hey certianly don’t predict that in 6 months time you will see a uptick in GDP as a result of putting more on the plastic today.
And what of the well documented long term trend for less traffic for around the same GDP output – how does your model cope with that?
Does it assume we have/are having a decline in GDP simply as less traffic is on the road?