Tag: public transport

Measuring Comparative Public Transport Accessibility for GB Cities

Posted in Mapping, Sustainability, transport, UncategorizedTagged , , , , , , , , , , Leave a comment

Although Greater London has an extensive transit network, this is not the case for many UK cities where underinvestment and privatisation has seen bus, metro and rail networks stagnate in recent decades, falling well behind European peers. Improving public transport is an important aspect of addressing the UK’s regional inequalities and poor productivity, and is a prominent issue for the 2024 general election.

Accessibility measures are an ideal tool to gauge the comprehensiveness and efficiency of public transport networks – they describe the ease with which populations can reach key services by different travel modes. The leading UK urban thinktank, the Centre for Cities (see their new Cities Outlook report 2024), has been doing some accessibility analysis of English cities compared to continental European cities, and this was recently republished in the Financial Times in an article on productivity challenges-

Centre for Cities report (left); Financial Time article (right).

It’s great to see accessibility analysis feature in the media. The measure used above however has some serious problems leading to nonsensical results (e.g. does Manchester really have half the accessibility of Liverpool and Newcastle?). The Centre for Cities measure uses a single time threshold (30 minutes) when we know that accessibility varies considerably at different time thresholds. It is based on a single destination point, when cities can have multiple employment centres. And it describes accessibility as a percentage of all city jobs, which means that the smaller the urban settlement is, the higher the accessibility result will be using this measure. In reality, larger city-regions have better jobs accessibility.

Creating Robust Public Transport Accessibility Measures – R5R and PTAI-2022
We can create much better and more reliable accessibility measures for UK cities. There have been significant recent advances. The open source R5R software has solved many of the computational challenges for accurately calculating public transport accessibility, allowing the calculation of full travel matrices for all possible trips and handling accessibility variation over time. In the UK, Rafael Verduzco and David McArthur at the Urban Big Data Centre have taken this one step further and pre-calculated accessibility indicators for all of Great Britain at a range of time thresholds in their Public Transport Accessibility Indicators dataset. This dataset is calculated using R5R, and is based on the median travel time across a three hour travel time window, 7am to 10am on a typical weekday (Tuesday 22nd November 2021), and uses the latest public transport service datasets such as the Bus Open Data Service. The results are at LSOA scale for GB only (no Northern Ireland), based on census 2011 zones (so I have used 2020 population data in the below analysis).

Origin and Destination Accessibility Measures
This article focuses on jobs accessibility, and this can be analysed from either the perspective of trip origins (residential-based accessibility to jobs) or from the perspective of trip destinations (workplace-based accessibility by residents). Both perspectives are complementary, and are developed below. For residential measures, if we take the average accessibility for all residents in a city then we get a good overview of how extensive and efficient the public transport network is. This requires city boundaries to define all the residents in each city. The analysis below uses the Primary Urban Area geography.

Public Transport Jobs Accessibility Trip Origin Results
The table and chart below show average accessibility to jobs for residents in all major GB cities by three travel time thresholds- 30 minutes, 45 minutes and 60 minutes. London’s accessibility results are inevitably much higher than any other GB city, being around 3 to 4 times higher at all three travel times, and emphasising just how big the gap is between the capital and all other GB cities. The 30 minute threshold describes shorter trips, and identifies higher density compact cities where residents are on average closer to employment centres. Small compact cities such as Cambridge and Oxford score well at 30mins (though note this is not the case at 45 or 60mins). Edinburgh and Glasgow have the highest residential average accessibility outside of London at both 30 and 45 minutes. This is due to Scottish cities historically following a higher density European urban model, and maintaining better public transport networks by avoiding some of the worst effects of privatisation.

The 60 minute accessibility measure picks up longer distance commuting on regional rail and metro networks. This is where the strengths of larger city regions such as Greater Manchester and the West Midlands are highlighted, with Manchester second and Birmingham forth in the ranking (Glasgow is third and also has a large regional rail network). Given their large populations, Manchester and Birmingham should however be scoring higher in absolute terms and closing the gap on London. Both have poor accessibility for the shorter 30 minute accessibility measure, reflecting the need for further inner-city densification (as the Centre for Cities have argued). For longer commutes, Manchester and Birmingham metro networks should also continue to be extended regionally. Leeds scores relatively well at 30 minutes due to its medium-density urban core, but it lacks a metro and is behind Birmingham, Glasgow and Manchester for the longer commuting times.

Peak Public Transport Accessibility by Trip Destination
We can also analyse accessibility by trip destination, which produces similar results to the trip origin residential measure but is more from the perspective of employment centres. The table below shows the peak accessibility by workplace within each Primary Urban Area, which is a measure of labour market size and agglomeration potential for the UK’s largest city centres. London retains its huge advantage with this measure, at 3 to 4 times higher than the next best cities. City-regions with larger rail and metro networks score better with the peak destination measure, with Birmingham and Manchester ranked second and third respectively, exceeding 2 million people at 60 minutes. Cities with strong rail connections to London, such as Reading and Crawley, also score highly at 60 minutes, but have much lower accessibility at 45 and 30 minutes. Smaller compact cities such as Edinburgh and Cambridge rank much lower by the destination measure compared to the residential analysis.

Both the trip origin residential average accessibility measure and the trip destination peak accessibility measure provide useful perspectives. The residential average measure is a good summary of the coverage and extent of public transport across a city, and how likely residents are to use public transport modes. The trip destination peak accessibility measures employment centre labour market size, and summarises the total number of people that can reach city centres by rail and metro. This is a better measure of agglomeration potential and is more closely correlated with city-region size.

Mapping the Accessibility Results
We can also map the results to view the geography of accessibility to jobs. Firstly the trip origin accessibility to jobs measure. This emphasises how large the area of high accessibility is across Greater London, with parts of Outer London and the South East having higher accessibility to jobs than residents in the city centres of the next largest cities, Manchester and Birmingham. The Primary Urban Area geography is also shown, which is the basis of the residential average accessibility chart and table shown above.

Next we map the trip destination accessibility to population measure. This has a very similar geography, but with more of an emphasis on city centres, as we are measuring average accessibility on a weekday 7am-10am when there will be more commuting services going to, rather than from, central areas. Again London has a huge advantage, peaking at 7 million people. We can also see the centres of Birmingham and Manchester reaching accessibility levels above 2 million people, while Glasgow, Leeds, Newcastle and Liverpool exceed 1 million.

Conclusion- Open Data and Software is Available to Create High Quality Accessibility Measures
With software such as R5R (see this workshop for an intro) and the exemplary and easy to use PTAI-2022 dataset from the UBDC, it is easier than ever to produce accurate public transport accessibility measures. The comparative accessibility analysis of GB cities shown here has highlighted the huge accessibility gap between London and all other UK cities. It has also shown the generally better accessibility performance of Glasgow and Edinburgh, and the high regional accessibility of Birmingham and Manchester which contrasts with their weaker accessibility in these regions for shorter travel times, which supports inner-city densification. There is no single perfect accessibility measure that answers all questions we are interested in – this analysis has confirmed that variation at different travel times reveals contrasting patterns in local and regional accessibility; that average and peak accessibility in cities emphasise different aspects of transit networks; and that trip origin and trip destination measures provide complementary perspectives. We therefore need to test a range of measures to understand accessibility patterns.

Future Improvements
This has been a relatively quick demonstration of the PTAI-2022 data and there are several areas for further improvements-

Including European cities for comparison would be very interesting, as the Centre for Cities explored in their original analysis. A recent major paper in Nature has shown how accurate international accessibility comparisons can be done- https://www.nature.com/articles/s42949-021-00020-2.

The PTAI-2022 dataset is a really good tool that makes GB accessibility analysis much more straightforward for researchers. Currently it uses the 2011 census boundaries, and the next update should use the 2021 boundaries allowing the latest census data to be used. Additionally, the current PTAI-2022 release uses 2021 public transport data, and updating this with the latest rail and bus data would also be a useful update. A related issue is that reliability on UK public transport networks can be poor, and that timetables can overestimate transit accessibility. This topic has been analysed by Tom Forth in this blog post.

This analysis has used the Primary Urban Area geography, which is a useful description of GB city-regions, but there are some issues with PUAs due to the underlying local authority geography. A few PUAs for medium-sized cities have quite large hinterlands (e.g. Sheffield) and this lowers the average accessibility measured in these PUAs due to lower accessibility outside of the urban core. A more thorough analysis of accessibility would need to test multiple urban geographies and gauge the extent of Modifiable Areal Unit Problem variation.

A Compact City for the Wealthy? Continuing Inner London Gentrification and Impacts on Accessibility Inequalities

Posted in City Planning, London, Mapping, SustainabilityTagged , , , , , , Leave a comment

We have a new paper out in the Journal of Transport Geography- “A compact city for the wealthy? Employment accessibility inequalities between occupational classes in the London metropolitan region 2011“. The paper explores how the increasingly affluent nature of Inner London has improved sustainable travel opportunities for more affluent professional and management classes, while less affluent groups have increasingly been priced out to lower accessibility Outer London locations.

The Continuing Gentrification of Inner London
The gentrification of Inner London was first recorded by Ruth Glass back in the 1960s, with middle class residents moving into largely working class neighbourhoods as London’s economy began its long evolution from manufacturing towards service jobs. This process has continued for decades, ultimately transforming most of Inner London. In the 21st century, some researchers have argued that gentrification has stalled (perhaps because there are few neighbourhoods left to gentrify) or has entered a different phase (e.g. processes such as super-gentrification and new-build gentrification as discussed by Davidson and Lees).

This research uses the Standard Occupational Class data as the basis of measuring social class. This classification was found to correspond to differences in income, as well as to a distinct residential geography. In particular, the three most affluent groups (Management, Professional and Associate Professional) cluster together, resulting in the social geography we can see in the map below using the 2011 Census data. There is a clear clustering of professional classes in Inner-West London, with two prominent radial corridors extending northwards through Camden, Islington and Hampstead; and south-westwards through Kensington, Wandsworth and Richmond. Concentrations of non-professional groups are mainly in Outer London to the east, north-east and west, with only smaller pockets remaining in Inner London. This analysis largely matches the description of Inner London now being dominated by professional classes, with lower income groups increasingly in Outer London (with some exceptions remaining in Inner East and South-East London).

Professional Classes (Manag., Prof. & Assoc. Prof.) Residential Percentage 2011. Data: Census 2011 (Office for National Statistics, 2016).

As well as mapping the 2011 geography of occupational class, we looked at more recent changes to see if gentrification is continuing or has slowed, using the ONS Annual Population Survey. Analysing changes between 2006 and 2016, we found had substantial gentrification had continued in Inner London, as shown in the table below. The Management, Professional and Associate Professional groups all grew as a proportion of the Inner London population, while all other occupational classes fell proportionally (green cells are above the average for the metro region, and orange cells are below the average). Interestingly, the biggest growth was in the Management and Professional classes, rather than the younger Associate Professional class, arguably more in line with super-gentrification processes. In contrast, there are proportional increases in several lower income classes in Outer London.

Sub-Regional Occupational Class Percentage Point Change by Residence 2006–2016 (final 2016 sub-regional percentages in brackets)

We can also explore these changes at the more detailed level of local authorities, and show that even more dramatic changes are occurring at the local level. In the chart below, each Local Authority is shown as a trajectory connecting its position in 2006 to its position in 2016 in relation to the percentage of professional classes and the total working population. Generally, Inner Greater London Authority (GLA) boroughs experience high working population growth combined with large increases in the proportion of professional classes. Boroughs with a long history of gentrification, such as Camden and Islington, are higher up in the chart reaching 70% professional classes, while more recent gentrifiers, such as Lewisham and Southwark, are rapidly gentrifying from a lower base. Outer GLA boroughs also show substantial population growth but with lower levels of change in professional classes, and decline in some cases. The exceptions are mainly in South and South-West London, with Croydon, Sutton and Richmond all gentrifying. Outer Metropolitan Area (OMA) local authorities have a mixed picture, with some increases in professional occupational classes with minimal working population growth; while some lower income towns such as Luton and Harlow are not gentrifying.

Local Authority Trajectories for Combined Percentage of Professional Occupational Classes and Total Working Population for 2006 and 2016. Data Source: Annual Population Survey 2005–2017.

What Impacts Does Inner London Gentrification have on Accessibility to Jobs?
We would expect that the dominance of more affluent classes in Inner London translates into accessibility advantages for these classes, as Inner London has substantially better accessibility opportunities by public transport, walking and cycling. We were particularly interested in accessibility by more affordable travel modes in this research. Bus travel is in general considerably cheaper than other public transport options in London. This is reflected in higher rates of more affordable bus and walking trips by lower income classes in the 2011 Census data. We can see in the table below that the three lowest income classes (6, 7 & 9) have around three times higher rates of bus travel and two times higher rates of walking than the most affluent three classes (1, 2 & 3)-

We used network analysis to analyse accessibility differences (see working paper on accessibility model). The analysis was carried out using the 2011 census data. The box plot below shows the cumulative accessibility to jobs for 60 minutes travel by Car, Public Transport (all modes) and Bus Only for the occupational classes. We can see differences between classes, particularly for public transport and bus trips, though there is also much variation within each class.

GLA 60mins Cumulative Accessibility to Employment by SOC Groups: Absolute Results

The accessibility differences between occupational classes can be more clearly seen by plotting differences between how the average accessibility for each group varies from the average accessibility for the entire working population, as shown below. Note in this chart the accessibility differences are normalised by travel mode, so the differences between travel modes in the chart above are normalised in the chart below. We can see clear consistent accessibility advantages for the top three occupational classes, particularly for more affordable slower modes- walking and cycling. The remaining occupational classes have below average accessibility to jobs, particularly for the more car oriented Skilled Trades and Process groups.

Greater London Authority 60mins Cumulative Accessibility to Employment by SOC Groups: Relative Differences in Occupational Class Mode Means and Mode Means for Total Population

The results for bus and walking modes is a particular accessibility challenge. Accessibility by these more affordable modes is generally low in absolute terms outside of Inner London. For the bus mode, less than half the number of jobs are reachable at typical commute times compared to the full public transport network. Given that lower income groups are the most frequent bus and walking commuters, and that these classes are increasingly being priced out of Inner London, these limitations are a significant accessibility challenge going forward.

What Policies Can Planner follow to Mitigate this?
In terms of transport policy, this research supports efforts to improve the affordability and connectivity of public transport for lower income populations. This is indeed a priority of the current London Mayor Sadiq Khan, who has committed to freezing public transport fares, and has reformed ticketing to allow multiple bus journeys on a single fare. These measures help offset travel costs for lower income residents in Outer London.

The main policy conclusion is the importance of housing policy in influencing accessibility outcomes in the study area. Low and moderate income groups are being priced out of public transport accessible areas. Without a step-change in the delivery of genuinely affordable housing in accessible locations, the increasing dominance of Inner London by professional classes will continue, resulting in greater accessibility inequalities, and likely increased travel costs for lower income classes.

Note on Covid-19 and Travel Inequalities
This research was completed in 2019, before the recent COVID-19 pandemic. The pandemic has in the short term shut down public transport networks, and greatly disadvantaged millions of city residents around the world. Longer term it is possible that the pandemic will reduce the attraction of inner city areas such as Inner London, due to perceived risk of future pandemics as well as the acceleration of telecommuting and home-working trends. The overall effect could be to slow gentrification processes, although this is difficult to predict. The alternative view is that  London will recover and adapt as it has done following many crises in the past. East Asian metropolises offer a good model of how to built resilience following their response to the earlier SARS and MERS outbreaks.

The wider economic impacts are clearly also important. Certainly we are in line for a very large recession, hitting important sectors such as tourism and hospitality. More specifically in London, the recession may hit development viability for affordable housing, and is a real headache for public transport operators. Transport for London was in financial trouble before the crisis, and is currently dependent on government bailouts to keep running. This will likely curtail the ability of the Mayor to maintain lower public transport fares, and so impact the kind of transport accessibility inequalities this paper discusses.

 

 

 

Open Source Public Transport Accessibility Modelling

Posted in UncategorizedTagged , , , , , , 4 Comments

The RGS-IBG annual conference has been on this week, and I presented as part of a series of geocomputation sessions arranged in advance of the 21st anniversary Geocomputation conference in Leeds next year. The topic was current CASA research from the RESOLUTION project, looking at developing fast and consistent methods of measuring public transport accessibility between different cities.

For this task I have been testing the OpenTripPlanner software with encouraging results. PDF of the slides are here.

The data used for the London analysis comes from the Traveline public transport timetable data. The image below shows an example accessibility measure of jobs accessible within 1 hour’s travel time leaving at 8am.

LondonResolution_PT_Accessibility