Summary: we are sharing a pre-print of analysis specifically looking at LTNs within Outer London’s mini-Holland schemes. Although the ‘LTN area’ sample size is small (most intervention areas weren’t LTNs) and uncertainty about effect sizes is large, we find consistent evidence about their direction. LTNs have reduced residents’ car ownership and/or use, and the already demonstrated increase in active travel from mini-Holland schemes is higher in LTNs.
Low Traffic Neighbourhoods have become a hot topic across the UK, especially in London. They form part of a series of relatively cheap and quick streetscape changes, which are being encouraged and/or funded by government as emergency measures to provide safer walking and cycling environments (safer from Covid-19 and from traffic injury risk) and relatedly, to try and discourage growth in car use given that public transport is still operating with major capacity restrictions. (See here and here for more).
What is a ‘Low Traffic Neighbourhood’? It’s a neighbourhood in which most or all through motor traffic has been removed from local residential streets (‘filtered’). This can be done in a range of ways: by planters, bollards, or other street furniture that physically block the road (emergency services can have key access to lockable bollards), by camera-enforced ‘gates’ (without physical restrictions, often so buses may get through, but fines may be imposed for illegitimate use), or opposed short sections of one-way street with cycle contraflows, intended to have a similar effect (less popular now, but some older schemes exist).
Low Traffic Neighbourhoods (LTNs), unlike many high-profile active travel interventions (such as cycle tracks, and pedestrianised city centres) target local areas – ordinary residential neighbourhoods where people live. They have multiple aims – most obviously, to make filtered residential streets truly quiet while still allowing residents, visitors, and deliveries to access all properties by motor vehicle.
This should benefit those living there: for instance, motor traffic levels are likely to fall below 100 vehicles per hour at peak, which according to design guidance in Manual for Streets, represents the upper limit above which ‘pedestrians treat the general path taken by motor vehicles as a ‘road’ to be crossed rather than as a space to occupy.’ In other words, Manual for Streets identifies a qualitative shift around that threshold, between a street being a place where people may wander, dawdle, or play even in the middle of the carriageway, to a movement corridor where pedestrians are relegated to their defined (often insufficient) footway space. Thus the aim of LTNs would be not just to reduce motor traffic by a certain %, but to change the character of those streets such that their entire width is usable by all, not just by those in motor vehicles.
LTNs are, however, not intended only to benefit residents or to increase ‘place’ function. They also aim to improve non-motorised connectivity. Within urban areas, a residential neighbourhood may be adjacent to key destinations, such as a park. An LTN in that neighbourhood will restrict people’s ability to drive through those filtered streets to the park. While this is an obvious benefit for residents (fewer cars cutting through), the filtered streets should themselves then form a much improved route by foot or by bike to the park, benefiting people living outside the neighbourhood, too.
This leads onto another posited benefit of LTNs – the discouragement of unnecessary car trips alongside the improvement of alternatives (principally walking and cycling, although where bus gates are involved, this gives buses a relative boost too). This is in line with evidence suggesting that just as building more roads tends to generate more motor traffic, reducing space for motor traffic will generally reduce motor traffic.
This isn’t magic: it can be as simple as someone realising their usual drive to the park is likely to take longer, looking at the map, realising it could be walkable and trying out the walk instead; discovering it’s quiet and pleasant, and now takes a similar amount of time to the car trip, so deciding to walk in future. ‘Modal shift’ achieved. We’ve seen millions of these kinds of small changes happen across London in the 1990s and 00s, when the major improvements in bus services and priority, and landmark pedestrianisation schemes, alongside an increase in the cost of driving and reallocation of roadspace away from the car, led to a substantial shift away from driving and towards public transport. Over two decades, London residents went from making around half (49%) of their trips by ‘private transport’ (mostly car), to making just over a third of them by ‘private transport’ (36%).
However, while we can identify large-scale change across London and other cities and point to the policy changes we think caused them, it’s often hard to link specific changes to individual schemes. Unfortunately, one does not get the whole picture just from monitoring changes in, for example, cycle traffic – this helps tell us what kind of routes are popular, but not how much of the change is new uptake nor where it came from. We need good evaluation methods to measure changes in travel behaviour but these methods aren’t cheap and their cost is rarely built into schemes up front.
I am lucky enough to have been leading the People and Places study, through which I have been able to look at the impacts of mini-Holland schemes in Outer London. Led by Westminster University and funded by TfL, the project involves a longitudinal study of adults age 16 and over, meaning that we follow the same people year-on-year. In this way we can compare changes in ‘intervention areas’ in mini-Holland boroughs to ‘control areas’ (the rest of Outer London). This controlled ‘natural experiment’ study design allows us to separate the effects of broader changes affecting all of Outer London (e.g. unusually good or bad weather) from the impacts of the programme. The study has focused on changes to walking and cycling (active travel) and has consistently found that living near interventions has led to a 40-45 minute weekly increase in active travel.
The People and Places study was not designed to evaluate LTNs. Some mini-Hollands interventions looked very much like LTNs, but others did not. Specifically, only one of three mini-Holland boroughs (Waltham Forest) implemented LTN-type schemes. Kingston and Enfield went for different interventions, generally more route-based; although Enfield has just recently implemented an LTN-type scheme in Bowes Park, and Kingston has now announced plans to introduce LTNs.
LTNs are controversial, as are many schemes that seek to restrict car use, and so we’ve been asked about our findings in this context. In particular, we have been asked what our findings mean in relation to ‘traffic evaporation’ – i.e. the extent to which measures like LTNs may lead to an overall reduction in motor traffic across an area. We had reported no statistically significant change in car use in our Wave 1 (2017) findings among residents of intervention areas, as published here. Amount of car use, like amounts of active travel, was measured in two ways: firstly, by any use in the past week, and secondly, by minutes using the car during the past week.
In more detail: ‘For past-week car use, there was a non-significant trend for those living in mini-Holland boroughs to be less likely to report any past-week car use than those living in non mini-Holland areas (p = 0.10). This trend was observed in all three mini-Holland boroughs, and the point estimate was somewhat stronger in the high-dose mini-Holland group than in the low-dose mini-Holland group although the differences were again not significant. Time spent driving in a car in the past week showed no consistent pattern in the results.’
So, does this mean that mini-Holland and/or LTN schemes don’t reduce driving? And can we say anything about how effective LTN schemes are, as opposed to other types of mini Holland scheme, in relation to active travel? With Anna Goodman, who’s worked on the project with me from the start, I’ve been doing some additional analyses of our dataset which address these questions more explicitly than anything we have published so far. We do so in all 3 years of follow-up data we now have available, Wave 1 (2017), Wave 2 (2018) and Wave 3 (2019), comparing these with the baseline year of 2016. Having three years’ follow-up provides us with more insight on trends – for instance, a weak trend that does not appear again is likely to be a blip, whereas a weak trend that replicates in a subsequent year is more likely to represent a real effect.
We have submitted an article entitled ‘Low Traffic Neighbourhoods, Car Use, and Active Travel: evidence from the People and Places survey of Outer London active travel interventions’ to a peer-reviewed journal, and uploaded a pre-print here. We have also uploaded a pre-print of our main three-year findings paper here.
That main findings paper discusses in more detail headline findings reported in TfL’s Travel in London 12, and uses Transport Appraisal Guidance methods to conduct an assessment of the physical activity and absenteeism related impacts of the schemes. It concludes:
‘These findings provide confidence that even in more car-dependent, suburban areas, active travel infrastructure can spur take-up, and that such growth can provide high health economic benefits in relation to intervention costs. Policy-makers should not however necessarily expect this take-up to immediately appear as increases in cycling; initially, active travel growth may manifest itself as increased local walking. Hence policy-makers should monitor changes in walking levels, which are often left uncounted; and (especially given controversy over cycling interventions) highlight the likely impacts of mini-Holland type interventions on walking.’
The LTN paper, by contrast, focuses on car use as well as creating a new subdivision: separating out LTN-type areas from other ‘high-dose’ intervention areas; which might be, for instance, near a new cycle track or a new pedestrian route, but not within an LTN area. We decided to look at car ownership as an outcome variable, which we haven’t done before. Having three years of follow-up means that we could better judge whether trends might be more likely due to chance; although note that there is plenty of uncertainty about the size of impacts (confidence intervals are wide) for the LTN group, as (depending on the wave) this is between 46 and 66 people.
Full detail of the results can be read in the papers, but to summarise the findings related to LTNs here:
In relation to car ownership, there is a consistent trend towards people in the LTN area being less likely to own a car, with the point estimate growing larger and more statistically significant in each subsequent wave (Rate Ratio (RR) 0.92, p=0.15 in Wave 1; RR 0.89, p=0.08 in Wave 2; RR 0.80, p=0.01 in Wave 3). Note that values less than 1 indicate a decline for the ‘%’ variables, while values less than 0 indicate a decline for the ‘minutes’ variable. A RR of 0.80 approximately corresponds to a 20% decrease in the adjusted probability of car ownership. By contrast, there was no evidence of any change in the ‘low-dose’ or ‘high dose, non-LTN’ groups.
In relation to whether a participant reported any past week car use, the largest decrease was again always the LTN group, with the effect significant in Wave 2 (RR 0.78, p=0.02) and borderline significant in Wave 3 (RR 0.81, p=0.08). There was a suggestion of a trend towards less past week car use in the high-dose, non-LTN group (RRs 0.93 to 0.96), but the point estimate was always smaller and never close to statistically significant.
As for minutes of past week car use, the point estimate in the LTN group was always negative (i.e. a decrease in time spent driving) and always lower than any of the other groups. The central point estimates were for a decrease of 10 minutes/week in Wave 1, 43 minutes/week in Wave 2 and 17 minutes/week in Wave 3. The confidence intervals were wide, however, and only in Wave 2, was the effect statistically significant (point estimate 43-minute decrease, p=0.007). That past week minutes of driving shows a decrease is notable as a concern sometimes raised about LTNs is that one may see some mode shift away from cars, but if all the remaining car journeys have to take more indirect routes on more congested roads then those car journeys will become longer and slower, and so the total volume of driving and pollution will go up. Our findings of a trend towards decreasing total weekly duration of car driving provide some evidence against this concern.
In summary, there was a consistent trend towards reduced car use in the LTN area for all three of these measures, with the effects particularly large in Waves 2 and 3. Confidence intervals were always wide, particularly for minutes of car use, but three of the six results from Waves 2 and 3 reached statistical significance and a further two were borderline significant. Despite small sample sizes, these results are therefore unlikely to be due to chance.
We also looked at our primary active travel outcomes in relation to these sub-groups. In general, in 16 of the 18 contrasts analysed, the point estimate was largest in the LTN area. Typically, the second largest was the high-dose, non-LTN area. Despite the small sample, many results for the LTN area reach statistical significance (4/6 for walking; 3/6 for cycling; 5/6 for active travel combined). The consistent pattern is therefore that one gets the largest active travel benefits in LTN areas, even larger than areas that still got a lot of high-quality new infrastructure but were not in an LTN.
It is also interesting to compare the magnitude of the estimated increases in active travel with the decreases described above for minutes of car use. The estimated increases in active travel are substantial in the LTN areas, albeit with wide confidence intervals, with point estimates of 94 extra minutes per week in Wave 1 (p=0.01), 67 minutes in Wave 2 (p=0.04), and 134 minutes/week in Wave 3 (p=0.006). By comparison the point estimates for decreases in car use were 10 minutes in Wave 1, 43 minutes in Wave 2 and 17 minutes in Wave 3. We cannot make definite statements about proportions given the wide confidence intervals, but the results seem most consistent with some of the increase in active travel reflecting mode shift away from car use, and some reflecting additional, brand-new walking and cycling trips or a mode shift away from public transport.