Commute pattern of 2.8 million workers in Hong Kong, visualised
What trends do the origin-destination matrix tell us about the supply of jobs in terms of location?
TL;DR
There are more than 2.8 million workers with a fixed place of work in Hong Kong. What does the overall travel pattern from their home to the working place look like, when visualised? The location of the jobs and the living place of the workers are revealing the job-housing imbalance problem of the new towns of Hong Kong.
From home to working place
Here’s an O-D matrix showing the number of workers grouped by their living and working place in 2016¹. Each cell represents the number of workers living in that row and working in that column. Say, the number 48,647 in the top-left corner means 48,647 workers are living and working in Central and Western District. The 12,938 next to it means 12,938 workers are living in Central and Western District while working in Wan Chai. On the other hand, the 19,244 below means 19,244 workers are living in Wan Chai while working in Central and Western District.
The boundaries used in the Census to classify the living place and working place of the workers are also shown below.
But the table cannot help bring some quick insights. People have to manually look up and compare the numbers one by one. How could we do better?
From table to heatmap
The following chart is called a heatmap. The vertical axis lists the districts where workers are living, and the horizontal axis lists the working location, classified by districts. Each square in the plot refers to the number of workers living in that row and working in that column. The darker the colour of the square, the more workers belongs to that living/working place group.
In other words, this chart is no different from the O-D matrix table above, except 1. the cells are now painted by a gradient-scaled colour scheme and 2. the numbers are stripped off.
But why colour it? Because we are more sensitive to colours instead of some number and digits. More importantly, the trends are easier to see with a gradient colour scheme. There could have various ways to interpret this diagram. I summarise 3 major trends and listed them below.
Trend I: The largest portion of workers works in the same district they live
One of the major trends you could see is the diagonal line of the heatmap has the darkest colour. This means the major portion of workers are living and working in the same district. Those living in Central and Western District are working somewhere around. And residents in Tuen Mun mostly work in Tuen Mun New Town. This is as expected — if possible, workers would find jobs near their living place to minimise the commute time.
The two exceptions are Sai Kung District and the Islands. In Sai Kung, a majority of the residents are working in Kwun Tung. Since Tseung Kwan O area falls into the Sai Kung District, we could expect most of the “living in Sai Kung, working in Kwan Tong” workers are living in Tseung Kwan O New Town (instead of Sai Kung Town and other rural areas). On Islands, as the Airport is classified as Other Areas in New Territories, all ground cabin for the airlines living in Tung Chung belongs to that grid. The Airport was a workplace for nearly 80,000 people (before COVID, of course).
Trend II: The jobs are still in the core urban area
The place of work belongs to Hong Kong Island and Kowloon (left-hand side of the chart) are in general much darker than those belongs to New Territories. With darker colour implies more workers, we know that majority of the jobs are still located in Hong Kong Island and Kowloon.
Trend III: New Territories → Urban Area flow is much larger than the urban area → New Territories flow
This is somehow an “extended” trend from Trend II. With more jobs located in the core urban area, it is easier for workers to find jobs there. A significant proportion of the New Town population thus works in the core urban area, making the lower-left corner of the heatmap (population living in New Territories and working in Hong Kong Island or Kowloon) quite dark.
On the other hand, those living in the core urban area are less likely to work in new towns. It is hard to imagine those 30 outliers living in Southern District while working in Tin Shui Wai New Town.
This is the phenomenon of “job-housing imbalance”/”job-housing mismatch”, an issue existed since the 1980s. The new towns were initially developed in a way that residents living in that new town would also work there. Looking into the master plan of Tuen Mun New Town, one could observe the east side of the Tuen Mun River is dedicated to residential development, while the west side is packed with industrial buildings. This light industry was supposed to be a major source of job positions for the residents.
However, the changes in the economic structure makes the jobs clustered around the two sides of Victoria Harbour. Millions of new town residents need to wake up early and take a one-hour wide from their home to the Central Business District.
Interested to know more?
I created a webpage (developed with shiny) that allows users to create custom charts specific to their area of interest with the O-D matrix data used in this article. The webpage is available below.
https://kenneth-12.shinyapps.io/place-of-work-od/
I am also planning to write a supplementary note about the webpage if possible.
[Update on 31 March: the supplementary note is available HERE]
[1] Note that workers without a fixed place of work or work from home are excluded.
