A map is worth a thousand insights

In the heart of Toronto's prestigious Bridle Path, November witnessed a staggering reality: the average home sale price surpassed a lofty C$4 million, translating to roughly a million dollars per bedroom.

This district, known in realtor parlance as 'C12', is one of the priciest neighbourhoods in Canada and home to celebrities like the singer Drake. A glance at the accompanying choropleth map, with District C12 conspicuously marked in red, offers a visual testimony to this economic stratification.

TRREB publishes these statistics each month in Market Watch. Yet, the data are exclusively presented as tables and that too in PDF format. This means one cannot process the data for further analysis, such as determining the average price per bedroom in each TRREB district. Furthermore, TRREB publishes a map in Market Watch that offers no statistical insights. The map is colour-coded to represent regions. It's a pretty map, but it's not insightful.

I have a history with the TRREB's district map. I digitized the map earlier in 1997 as part of research for my Master's thesis. Because of the sales data TRREB shared, I developed valuation models and embarked on a lifetime journey of studying the spatiotemporal evolution of real estate prices. The map below is the first time TRREB's data were presented in a thematic map. I know this because I spent two years geocoding half a million sales records that had not been processed by TRREB or anyone else.

With Geographic Information Systems (GIS) in its infancy and street network files being either unavailable or patchy, assigning longitude and latitude to each transaction (geocoding) was no small feat. However, I was motivated, and even obsessed, to understand how buyers and sellers agree on a sales price. Spending an extra year to geocode the sales data was a small price to pay to be the first one to generate spatial insights.

My passion for applying a spatial lens to housing markets has not faded in 25 years. In fact, I have been working on updating the GIS version of the TRREB map. The map's boundaries have changed since 1997. I started working on this task last year with my research team at the Urban Analytics Institute. We made good progress, but getting the boundaries right is no easy task.

TRREB does not readily make a digital map version available to researchers. So we tried to guesstimate, approximate, and use other 'scientific' approximations to produce version 1 of the digitized map corresponding to the tabular data TRREB publishes in PDF format. Our colleagues at Caliper Corporation supported us with ideas and algorithms to digitize the map using Maptitude software.

Isn't it odd that the world has embarked on big data and AI while Canadian researchers struggle to access real estate transaction data?

With a GIS representation of the map, we can merge data from the Census and other sources to generate new insights, such as determining the average price per bedroom in the GTA. Note that our first map above plots the average price, which gives the impression that housing in downtown is priced the same as in the outer suburbs like Oshawa. However, such a naive comparison masks the fact that downtown Toronto sales are mostly small-sized condominiums with one bedroom. In contrast, suburban sales are mostly low-rise housing with three or more bedrooms on average.

When we plot the average price per bedroom, a different picture emerges. The downtown condos are in the same category as the dwellings in the Bridle Path, where the average price per bedroom exceeds half a million dollars. To the southwest of the GTA, Oakville is another jurisdiction with high-priced dwellings- easy to spot, given the red-coloured theme.

A map is worth a thousand insights.

Unlike the United States, real estate data are not readily available, in fact not at all available, to researchers in Canada. Isn't it odd that the world has embarked on big data and AI while Canadian researchers struggle to access real estate transaction data?

They say data is the new oil. Without data, it is next to impossible to do meaningful research on housing markets. No wonder while taxpayers pay the salaries of academics and researchers in Canada, the research they produce is often focused on American housing markets. Why? Because American housing sales data are readily available.

Housing affordability is the most pressing problem in Canada. The next federal and numerous provincial elections will be contested in the coming years on this singular issue. Despite the urgency and currency, housing sales data have remained under lock and key, denying Canadians the resources they need to generate solutions for housing affordability.

It doesn't help that even the data collected by Statistics Canada is not readily accessible. The Canadian Housing Statistics Program (CHSP) is an initiative of Statistics Canada that "leverages existing data sources and transforms them into new and timely indicators on Canadian housing." CHSP has been the source of much-needed insights that have helped inform debate and discourse on housing affordability in Canada. For instance, CHSP helped answer how widespread foreign home ownership was in some select provinces.

To access CHSP data, one must go through some challenging hoops and cope with logistics that were impossible during COVID-19. First, one must write a proposal to advise why one needs access to CHSP data. Then, one must undergo security clearance and other formalities. Finally, the data are available at Statistics Canada's designated Research Data Centres, such as the University of Toronto and McMaster University.

My team and I were the early users of the CHSP data. However, the pandemic disrupted our research when we couldn't access the data centres for months/years. When physical access to data centres was finally restored as the pandemic became manageable, the team had already moved to other projects, and our efforts bore little fruit.

To foster a robust, data-driven approach to housing affordability in Canada, it's imperative to democratize access to housing transaction data. Academics, researchers, and data specialists must be empowered with the tools to analyze and propose solutions to this pressing national issue. The time for action is now – to unlock the potential of data and chart a course towards a more affordable housing landscape in Canada.

Regionomics Inc. is a Canadian consultancy specializing in applying predictive analytics and machine learning to find solutions for economic challenges. For more information, please contact Murtaza Haider at [email protected].