In this recent post Matt mapped census journey to work data across Auckland.
I will now use this data to draw some high-level inferences about returns on Goverment investment in transport infrastructure in Auckland in the period from 2007-13, i.e. between the last two census. The specific question I want to answer is: Given the level of Government expenditure in different transport modes in Auckland in the last 7 years, which mode(s) appear to have delivered the best returns in terms of their ability to accommodate growth in journey-to-work (JTW) travel demands?
Let’s look at the data. Total number of JTW trips by mode for the last two census are summarised in the table below, along with the change (growth) between census.
- Note 1: “Private vehicle” combines JTW trips for private/company cars, trucks, passengers, and motorcycles.
- Note 2: Public transport combines JTW trips for bus and rail.
This shows how an additional 23,223 JTW trips were undertaken on census day. Of these trips, 10,200 (44%) were undertaken by car; 9,118 (40%) were undertaken by public transport; and 3,825 (16%) were undertaken by walking/cycling.
Matt has helped pull together a spreadsheet of government transport investment by mode from 2006-13. This includes investment by NZTA (nee Transit) and local government. In the table below I have added total expenditure by mode, and then used this to calculate an average cost per (new) JTW trip and (additional) JTW-km.
Note 3: Average trip length was calculated as the weighted average trip length using HTS data.
Now we start to see some interesting trends. From these numbers it appears that Government investment in public transport and walking/cycling has been approximately 2.6 and 6.7 times more cost effective than investment in roads during this period, at least when measured on a per JTW-km basis (the differential is even greater when you use the $ per JTW trip indicator). This in turn suggests that it would be more cost-effectively to cater for growth in JTW travel demands by shifting investment away from roads and into public transport and walking/cycling.
This raises the question of how much should our current investment levels shift in order to deliver approximately neutral investment outcomes for each mode? To help with this I have calculated what I call a “neutrality factor”. For each mode this is calculated as the ratio of 1) it’s share of growth in JTW-kms divided by 2) its share of total expenditure. This tells you how much expenditure on a particular mode would have to change to have achieved a ratio of 1.0, i.e. a mode’s share of JTW-km growth was equal to its share of Government investment.
In the table below I have calculated the neutrality factor for each mode.
When calculated on this basis, it would suggest that we should increase government investment in public transport and walking/cycling expenditure approximately two-fold and five-fold respectively, while reducing investment in roads by approximately 30%.
Of course, this analysis is a “partial analysis” because it assumes demand (number of new JTW trips and kms) is held constant. In fact, if we were to see such a shift in transport investment priorities over a sustained period of several years then you would expect the demand for each mode to respond accordingly. More specifically, shifting government investment away from roads and into public transport and walking/cycling would be likely to encourage more people to use those modes, which would in turn warrant greater government investment. Hence, the changes in transport investment priorities suggested by this analysis are likely to be relatively conservative.
This is obviously a back of the envelope (“BOE”) analysis.
The most obvious issue is that JTW trips account for only a proportion of total travel. Education travel is another major contributor to peak period congestion. However this tends to be slanted in favour of public transport and walking/cycling – such that including these trips in the analysis seems likely to confirm these outcomes.
The other is that we are lumping together capital and operating costs. While this makes sense in the long run (because after all, capital investment is really just a very lumpy operating cost!) it may be distorted in the short run. Hence it might be useful to extend our analysis by calculating the effectiveness of transport investment from 2001 to 2013. Indeed, the period from 2001-2006 seems to have been associated with less dramatic changes in travel demands than the period from 2006-2013.
While this analysis certainly could be improved, even now it seems to provide a useful indication of the relative returns from Government investment in different transport modes from 2006-13. I also feel somewhat justified using it because members of the current Government frequently justify their transport investment priorities on the grounds of raw 2006 JTW mode share. I’ve followed their lead and focused on JTW trips, but instead looked at the change in demand over time and then compared it to relative expenditure levels.
This tells quite a different story than if you look at raw JTW mode share.
On this basis I’d suggest that Government investment in public transport and walking/cycling in Auckland has typically delivered more cost-effective returns than investment in road based transport infrastructure. So the next someone tries to argue that we should invest more in roads because “86% of Aucklanders drive to work by car”, please challenge them on the basis that relative potential for growth – rather than the level of existing demands – should be the primary determinant of our future investment priorities.
Footnote: In the most recent census private vehicle mode share appears to have dropped from 86% in 2006 to 83% in 2013. Given that we’re yet to see the effects of electrification, new EMUs, integrated ticketing, and the New Network, one would have to think that the relative returns from Government investment in public transport might be similar high if not higher by the time the next census rolls around.
Processing...
Fascinating, Stu. Do you have the data to hand to run the same methodology for all trips, or at least to include education trips as well? I note that the Australian Census asks this question as ‘Journey to Work or Education.’ This would reflect user demand much more accurately as we know students this modes differently than workers.
Census only captures journeys to work. Maybe other surveys would look at different trips but they’d clearly have WAY smaller sample sizes.
Exactly. Why don’t we collect this data too? Getting to University or school is the work equivalent for young people and is something that we want to be as efficient and safe as JTW. This should be added as a separate question identical to the JTW one so the data can still be analyised on its own in order to still be able track changes over time with the JTW answers from earlier Census.
while working at NSCC I did a comparison of JTW to the CBD by bus with the annual Fanshawe St cordon count, apples and pineapples to be sure, particularly as it made the heroic assumption that JTW trips were roughly peak period trips, but it did suggest that around a third of AM peak bus trips into the city were for non-work (most likely study) purposes
would be great to have a question on education travel added to the census 🙂
I think that you will struggle to get a change in question as the cenus people like to keep the same question so that they can have continuous data. I’m sure Stats NZ could provide more detail but fear that is the case
They could simply add a question, repeat it but change ‘work’ for ‘education’
I’ve expressed my thoughts on this in some previous comment, but my dissenting view is that it should be left the way it is; both for comparability, and because there are other, easier ways of getting the data (i.e. talk to the universities direct). The census always has to strike a balance between getting as much useful info as possible and minimising the burden on the people filling out the forms (and the risk that people look at the form and go, this is too hard, I can’t be bothered filling it out at all).
Well OK JP, if we have good data from other sources let’s run this analysis including Journey to Education too. Clearly the result will further highlight the hopelessness of current policy; disproportionate spending on the one complete network for lower return and the failure to direct investment to the more productive areas of the sector.
Or how about replace it with two questions:
For the longest distance journey you undertook today, what was its primary purpose to get to or from?
(Work/Education/Social/Other etc)
What mode(s) of transport did you use on that trip? Select all that apply:
(Private vehicle/Bus/Train/Ferry/Walk/Cycle etc)
I agree with that sentiment absolutely Patrick!
The current question asks how you travelled to work, regardless of whether you studied (I did both on Census Day 2006, for example).
If you added questions: “did you travel to study and if so how”, they wouldn’t add any burden to people already answering the work question (since they’d skip it). You’d also make sure the data was comparable with previous years. If you want to compare with historic data, use only the answers from people who said that they worked.
The only thing lost there is the travel-to-study for people who both study and work, if people make a big trip for study, but not for work, since only the little trip would be counted.
I’d suggest we do a second question following the JTW one as follows:
[remember JTW question is **only** answerable for those who in the 7 days ending the Sunday prior to Census day worked at any job in return for money]. Anyone else can’t answer the JTW question.
“Did you also travel for any non-work related purpose today? – (non-work related means it was not related to your employment and also excludes travelling to or from your place of employment)”
If so please indicate how you undertook this journey – choose only the longest part of the journey if you used multiple trips or transport methods on this journey: Select: {Private vehicle/bus etc etc).
The main purpose of this journey was: {Education e.g. School or University/Study/Shopping/Social/Other} and this journey was approximately XXX km “one way”.
This would capture the main non-JTW journey for everyone who did one, and also how they did it and approximately how far they went.
You could then ignore the non-JTW answers to get the consistent set for JTW for comparison and also have a second set of data which indicates the types of non-work related journeys being undertaken.
You won’t know the destination for the non-JTW one – only the approximate distance covered but that is better than the nothing we have now.
You could also see how many people travel to work and don’t do any other travelling.
Steve: the thing is, there are lots of interesting questions that could be added to the census, but difficult choices need to be made as to what gets included. They’re selected to give a broad coverage of a range of social, economic and demographic variables. Only 15% of the over-15 population are engaged in study, so you’re adding in an extra question which only gets answered by a small proportion of census respondents, and not giving us that much info which we don’t already know; we know that students are big users of PT, and this can be tracked fairly accurately because most of them presumably make use of the sizeable discounts they get on their trips. By comparison, most census info is hard to collect any other way, which is a key reason for asking those questions in the census.
Greg: There are various ways to analyse how much travel people do in private or public transport, including the Household Travel Survey etc (which is probably closest to the kind of questions you’re suggesting, and is carried out on an ongoing basis), HOP data (I’m a bit surprised we haven’t seen much data released from this yet), the NZTA’s highway volume indicators and MoT’s aggregate vehicle travel stats which we publish on the blog from time to time, Household Economic Survey data, surveys, and so on. I’m giving a presentation at a conference next month which will look at another data source.
The Household Travel Survey is quite interesting and I’ll try to write about it some more in the future. I did a guest post back in the day, here: http://greaterakl.wpengine.com/2013/01/07/pt-frequency-of-usage/
Sorry for being a stickler about the census questions, but I spend a lot of time working with this stuff, and I’m simply trying to show that there are valid reasons for not adding extra travel questions to the census – even though they would no doubt give interesting results.
John – all your comments on not having more questions in the Census are valid, and come with one. big. BUT.
And that is, by asking those extra questions at Census time you get a consistent set of data, that is aligned to Census JTW answers 100% as its taken at the same time.
Taking HTS or other data (University data or HOP data) never aligns with the census and doesn’t have the depth of information the Census does.
So there is always a gap between what the HTS shows and what the Census shows. While HOP data is useful its only capturing information 200,000 (20% at best, more like 15%) of the population.
And only where they got on or off the PT system – not why they are travelling, where they work, or where they live.
Secondly, the last Census showed that the on-line proportion of respondents is growing (25% if I recall).
So its not that much more of an imposition to have a few more questions if a fair chunk if not most of the respondents are on line and are thus not really going to be “disadvantaged” by answering them – something it will take them 30-60 seconds longer to answer. And it gives a consistent data set you cannot get any other way.
Census questions come and go – they no longer ask about how many Chickens you keep, or how many domestic servants, or what home appliances you have now do they? They used to right?
And every question in a census had to be asked for the first (and last) time at some point – all I’m proposing to start a second travel related set of questions to capture those who are NOT in paid employment at the time of the census – as after all their travel needs are as important as those who are in paid work and have just as big a part to play in daily travel congestion
– witness how much better the roads are in the term school holidays. Thats mostly the “Education” traveller – and almost NONE of those people’s travel counts in the Census. And its simply time it did. We owe it to ourselves to get better informed. The existing JTW questions can stay as they are for sure, But we need some more.
If you disagree, then can you show me how the existing data sets align with the Census data to provide a consistent set of answers – because I don’t believe they do.
I coordinated an effort to have travel to education included in the 2001 Census, Stats have masses of data to collate and have to look at the big picture, so a solution that entailed the least amoungt of change was the only way to go
we managed to get most major city and regional councils to submit, but were (I think) sunk by some councils not understanding Stats position and requesting massive detail which would have had marginal usefulness
maybe with Auckland Council representing a third of the country’s population and with the greater automation of data collection, it’s time to have another try
Greg, for me, it just comes down to what’s in the first paragraph of my comment: the need to weigh up which questions to include, out of a vast number which could be asked on all sorts of topics. Taken together, the various data sources give a pretty good picture of travel information, even if they all measure slightly different things.
“shifting government investment away from roads and into public transport and walking/cycling would be likely to encourage more people to use those modes, which would in turn warrant greater government investment.”
So the reverse applies: the relative failure would have been even worse without highly-funded encouragement to use roads during recent years.
yes!
Great to see this sort of analysis.
Since most of my comments have already been dealt with in the comments, I’ll just add two observations.
I’m surprised that the total number using PT is so small. When you think of all the trains, buses, and ferries out there, it seems like it should be more than 43,000.
And, it would be interesting to see journey to shop, journey taking the kids to school, and other such things. I would bet that’s a much higher proportion of automobile trips.
Hi Stu. I have been thinking about your post all day and I haven’t read the comments so apologies if this repeats. First I am not sure you need to consider trip length at all. Simply taking your cost per JTW and calculating an effectiveness in terms of private travel. That increases PT to an effectiveness index of 2.8 but active travel up to 33. My point being if you want value for money you want to get people to work, how far they go doesn’t need to be an input. Particularly if you can get people to walk a short distance you are scoring a real efficiency gain. It should be trips we compare because the trip brings benefit not the distance someone has to go. Second I havent gotten my head around the argument you build in the last table. But I think you can get straight to a funding neutrality argument straight off effectiveness and again ditch the veh-km given that increasing veh-km is an observed outcome but not actually an objective or something to be maximised or controlled. That said each time I read the 2nd table I like it more. It is simple and makes the point.
yes I tend to agree with you: The per trip costs are more relevant.
I used the per-km rate because it understates the case, if anything.
“So the next someone tries to argue that we should invest more in roads because “86% of Aucklanders drive to work by car”, please challenge them on the basis that relative potential for growth – rather than the level of existing demands”
PT is growing at 0.212% per year over the last 17 years – hardly a great argument to annoy the 86% of Aucklanders driving cars.
Yes, the potential for growth should be the primary driver of investment, not the current “value”.
Arguing we should invest in roads because they currently have high mode share is like arguing people should buy Apple shares because their share price is currently high.
QED.