Is it better to buy or rent? - Methodology

This page describes the methodology used in our interactive story Is it better to buy or rent? which was published on October 8, 2026. The code used for the calculations is open source and available within the journalism TypeScript library (journalism library, simulation logic, monte carlo engine).

For questions and comments, reach out to CBC News Senior Data Producer Nael Shiab.

Summary

You don’t have a choice: you need a place to live. But should you take on the steep costs of homeownership, or rent a cheaper place and invest the difference? Which option leaves you with more money after 25 years, once you sell everything off?

To find out, we built a simulator using over two decades of historical data across 22 Canadian metropolitan areas, covering mortgage interest rates, maintenance costs, property taxes, rents, and more. Our premise is simple: every month the homeowner pays more than the renter, we assume the renter pockets those savings and invests them directly into the stock market.

We ran this scenario thousands of times and here's what we found:

• If you are willing to sacrifice space and live in an average studio or one-bedroom, renting is more financially advantageous than owning an average condo. But to come out ahead, you would need to invest in an aggressive portfolio, gambling on the stock market's volatile ups and downs.
• However, if you rent an average two-bedroom or bigger, the math flips. The average condo becomes a better option, whether you are an aggressive investor or a cautious one.

In other words, buying a condo is often the "better" option simply because owners lose less money than renters do.

This is not true for house owners. Most of the time, the growing value of their property outpaced its accumulated expenses, leaving them with a true profit at the end of 25 years.

Ultimately, though, the "buy vs. rent" debate is purely theoretical for most Canadians. Half of all couples don't earn enough to qualify for an average condo in their hometown, and that jumps to two-thirds for a house. Most Canadians must rent, even though renting is often the least advantageous financial option, especially for families needing an extra bedroom.

Run the simulation for yourself with your own parameters below.

Methodology overview

We built a Monte Carlo simulation to compare buying versus renting in Canada over the next 25 years. Based on historical data for 22 major metropolitan areas, the model uses the volatility of home prices, rents, interest rates, and more as a guide to generate hundreds of statistically plausible scenarios that provide insights into the potential financial impacts of each choice.

In our simulation, we assume that when a renter’s monthly costs are lower than a buyer’s, the renter invests the difference in the stock market. This means that if an owner's expenses for a given month are $2,000, while a renter's are $1,800, we assume the renter invests the $200 difference. However, if the renter has higher monthly expenses, they have nothing to invest.

For the buyer, the only investment considered is their home. Our model contrasts the financial outcomes of a renter with those of two types of homeowners: one with a fixed-rate mortgage and another with a variable-rate mortgage, both renewed every five years.

To make our calculations as accurate as possible for the Canadian context, we account for mortgage insurance, land transfer taxes, property taxes, Tax-Free Savings Accounts (TFSAs), sales taxes, and capital gains taxes, when applicable. We also adjust all monetary values for inflation, so they are shown in today's dollars. Our baseline simulations assume the household is a couple.

Other interesting findings

• Variable-rate mortgages performed better than fixed-rate mortgages in the majority of our simulations. Variable-rate buyers often ended up with a larger profit (or a smaller loss) than fixed-rate buyers. The difference is not always substantial, but it is interesting to note. In some cases, the variable-rate buyer is even able to finish paying off their mortgage before the 25-year period ends.
• What is counterintuitive to most is that, when adjusted for inflation, homeowners' monthly expenses tend to decrease over time. This is because the biggest portion of their costs (the mortgage payment) is not tied to inflation. During their term, the payments stay the same. When the mortgage is renewed, it is based on the remaining portion of the mortgage, which is not reassessed. Renters, on the other hand, usually see rent increases that exceed general inflation. This means renters' real expenses usually increase over time, while owners' decrease. This dynamic is a key reason why renters can end up further behind buyers over 25 years: eventually, being a renter can cost more per month than being an owner, even if it started out cheaper. In our simulation, this means renters are not saving any more money per month and stop investing, while owners continue to build equity in their property.

Overall simulation results

Simulator

The simulator works by generating multiple statistically plausible futures for the expenses and investments of renters and buyers over a 25-year period. After each iteration, a winner is determined based on the final wealth of the renters compared to the buyers.

Our model tests three scenarios:

• A homebuyer with a fixed-rate mortgage, renewed every five years, over 25 years
• A homebuyer with a variable-rate mortgage, renewed every five years, over 25 years
• A renter staying in the same apartment for 25 years

In our modeling, we assume that the household is the same across all three scenarios and is always able to pay for the most expensive option. Any savings made by the renters compared to the owners are assumed to be invested in the stock market. If the renters have the highest monthly costs, they have nothing to invest. For the owners, their property is their only investment in our simulation.

The user can test the simulator by adjusting the parameters in the interactive tool below. Detailed explanations of the calculations, assumptions, and limitations of our model are presented after the simulator and results sections.

Configure the parameters above and run the simulation to see results.

Side-by-side scenario comparison

Configure a second scenario below and compare its results against the simulation above. The table shows median assets, selling costs, and total after selling at five-year intervals over the 25-year period.

Assumptions and limitations

We assume that the household is the same across all three scenarios (buyer with fixed-rate mortgage, buyer with variable-rate mortgage, and renter) and is always able to pay for the most expensive option.

By default, only the renter is able to invest the difference between their monthly expenses and the homeowner's monthly expenses. This assumes perfect financial discipline for the household, which, of course, is not always the case in real life.

We also assume this is the household's primary residence and they will live in it for the entire period. For buyers, this means the proceeds from the sale of the property will not be taxable. For renters, any gains from the sale of investments in unregistered accounts will be subject to capital gains tax. Of course, in real life, people often move and may have different tax situations.

Note that our model takes TFSAs into account and prioritizes them by default. Since future government limits for TFSAs are not yet known, the model assumes the annual contribution room will continue to increase by approximately 4.2% per year, which is the historical average since the TFSA was introduced in 2009. Once the TFSA limit is reached, any additional savings automatically flow into a taxable stock portfolio. In our model, we assume the household starts with a $0 TFSA balance. Consequently, capital gains tax is almost never applied in our simulations as most investments stay within the tax-sheltered limits.

We do not include other tax-advantaged accounts like RRSPs in this simulation. This is because RRSP contribution room is highly individualized, as it is based on the user's personal income history. Additionally, our model's approach of simulating a total exit at the end of 25 years would trigger a massive and unrealistic tax bill for an RRSP, which is typically designed for gradual withdrawals during retirement.

Specific calculations for mortgage insurance, sales tax, land transfer tax, and capital gains tax are based on 2024-2026 rates and regulations. For land transfer tax, our model automatically calculates the cost based on the metropolitan area and property value. It also accounts for first-time homebuyer rebates where applicable, assuming the buyer meets the standard provincial or municipal eligibility criteria.

Our analysis is limited to the financial aspects of buying versus renting. Other important factors of owning or renting a home are not captured in our model.

Another important limitation is the difficulty of establishing a direct equivalence between rental and owned properties. This is due to how historical data is categorized by different institutions:

• Rents (CMHC): Categorized strictly by the number of bedrooms (Studio, 1-bedroom, 2-bedroom, etc.).
• Property prices (CREA): Categorized by building type (Single-family, Townhouse, Apartment/Condo, etc.) regardless of the number of bedrooms.

As a result, while our default simulation compares a "two-bedroom apartment" for renters with an "apartment" for buyers, the historical benchmark for the latter includes condos of all sizes (studios, 1-bedroom, 2-bedroom, etc.). This mismatch is an unavoidable consequence of relying on long-term historical data sets (spanning more than 20 years) in Canada and should be kept in mind when interpreting the results. For these reasons, utilities are also difficult to estimate and are not included in our model.

Finally, it is important to note that the historical data used in our simulations, including housing prices, generally covers the last 20 to 25 years. During this period, Canada did not experience a major, long-term housing crisis. The last such event occurred in the 1990s and is not reflected in our dataset. As a result, the extreme risks and outcomes associated with a deep housing crash are not captured in our analysis or our simulator.

Explanation of the calculations

Starting parameters

The model requires a set of starting parameters for the year 2026. In the simulator above, they are hidden by default but the user can click "Show parameters" to view and adjust them.

These parameters are increased or decreased over time, based on the generated economic paths. Here is a description of the parameters for the default values used in the simulator (one-bedroom apartment in Montreal).

For the renter:

For the buyer:

We also need the following values to calculate the cost of selling the assets and trading fees:

Historical data and modeling

To provide a realistic foundation for our calculations, we rely on up to 25 years of historical data. This historical record is used to build the statistical models that generate thousands of plausible future paths.

Specifically, we analyze this data to determine the historical volatility and drift of each economic variable. These parameters are then used in stochastic models: Geometric Brownian Motion (GBM) for most variables and the Cox-Ingersoll-Ross (CIR) model for interest rates. GBM is widely used for modeling stock prices, home prices, and rents because it accounts for long-term growth and random fluctuations while ensuring that values never drop below zero. The CIR model is particularly important for interest rates because it is "mean-reverting," meaning it reflects how rates tend to return to a long-term average over time, preventing them from becoming negative or rising indefinitely.

Note that the historical record length varies by variable. While investment portfolios, interest rates, and rents are analyzed from 2001 onwards, the benchmark home price data from the Canadian Real Estate Association (CREA) begins in 2005 for all metropolitan areas.

Correlated economic paths

A crucial feature of our simulation is that economic variables do not move independently. In the real world, factors like interest rates, home prices, and inflation are deeply interconnected. To capture these relationships, our model is powered by 1,760 bespoke Cholesky decomposition matrices, pre-computed for every possible combination of metropolitan area, property type, rent type, and investment portfolio using historical data from 2005 to 2025.

By applying the specific matrix corresponding to the user's scenario, we ensure that the fluctuations generated during the simulation follow the unique historical patterns of that local market. For example, if a scenario randomly generates a spike in inflation, other related variables—like insurance costs, maintenance, and property taxes—are statistically more likely to rise as well, following the specific economic "rhythm" of the chosen metropolitan area and scenario. This provides a highly localized and realistic assessment of the risks and rewards of buying versus renting.

Investment portfolios

Stock market portfolios are built using historical performance data for three core indices: Canadian bonds (iShares XBB), Canadian large-cap stocks (iShares XIU), and U.S. large-cap stocks (SPDR S&P 500). All U.S. returns are automatically converted into Canadian dollars based on the exchange rate of the month, as reported by the Board of Governors of the Federal Reserve System (U.S.).

Our model uses four different investment portfolios to represent various levels of risk:

The values in the chart below are normalized, so all portfolios start at 1 in January 2001. These values are used to create the parameters of the Geometric Brownian Motion (GBM) model that generates the future paths of investment returns in our simulation.

Mortgage and interest rates

The interest rate data shown in the chart below represents the posted rates from the Bank of Canada. These are the official rates announced by banks, which are typically higher than the actual rates negotiated by consumers.

Data for the actual rates paid by Canadians is only available from the Bank of Canada starting in 2017. To accurately model how rates fluctuate over a longer 25-year period, we used the posted rates (available since 2001) but added fixed (-1.99%) and variable (-0.39%) adjustment parameters to our simulation. These parameters are based on the average difference between posted and actual rates observed from 2017 to 2026. In all cases, a minimum interest rate of 1% is maintained. You can see the adjusted rates in the chart below.

Additionally, since the Bank of Canada does not provide posted rates for two-year and four-year fixed mortgages, we have linearly interpolated these values from the available one-year, three-year, and five-year data. These additional terms are important for calculating mortgage penalties.

These values are used to create the parameters for the Cox-Ingersoll-Ross (CIR) model, which generates the future paths of interest rates in our simulation.

Property prices

Home prices are based on the Canadian Real Estate Association (CREA) MLS Home Price Index. This data allows us to track the benchmark price of different types of properties across major Canadian metropolitan areas. Note that this dataset begins in 2005 for all metropolitan areas, as benchmark data starting in 2001 was not available.

These values are used to create the parameters for the Geometric Brownian Motion (GBM) model, which generates the future paths of property prices in our simulation.

Rent prices

Our model uses two sets of rent data: one for historical modeling and one for the starting rent of the simulation.

Historical rent data

Historical rents are based on a hybrid calculation that combines data from the Canada Mortgage and Housing Corporation (CMHC) and Statistics Canada’s Consumer Price Index (CPI) for rent. These two sources were chosen because they provide the most comprehensive and long-term data on rent prices in Canada, despite having different methodologies and scopes. It's the closest we can get to an estimate of how the cost of occupied rental dwellings (which is what renters pay) has evolved over time across the country.

Our methodology uses the CMHC rent from 2001 as a starting point and projects it forward using provincial CPI rent indices. However, since the CMHC (which focuses primarily on the primary rental market) and the CPI (which measures a broader universe of occupied dwellings) often diverge, we calculate the average of the CMHC-measured rent and the CPI-adjusted rent. This provides a "middle ground" value that accounts for both specific market measurements and broader inflationary trends.

This hybrid approach was chosen to maintain consistency across all provinces and metropolitan areas. Although each province has its own rent control regulations, our model uses these standardized sources to ensure that comparisons between different regions are based on a uniform and reliable methodology.

These values are used to establish the parameters for the Geometric Brownian Motion (GBM) model, which generates the future paths of rent prices in our simulation.

Starting rent values

For the starting rent values used in the simulation, we could not rely on the CPI because it is an indexed value that does not focus on vacant dwellings. Similarly, we could not rely solely on the CMHC because it targets the primary rental market. Therefore, we opted to average the CMHC data for vacant dwellings (recently available) and the Rentals.ca data, which covers both primary and secondary rental markets.

In some cases, rents for vacant dwellings were not available in the CMHC data, so we used the overall average rent as a proxy. When certain data points were missing from Rentals.ca, we stuck to the CMHC values.

Consumer Price Index (CPI) variables

To project how recurring costs like insurance, maintenance, and property taxes might change over time, our model uses specific categories from the Consumer Price Index (CPI) provided by Statistics Canada.

The values described below are used to create the parameters for the Geometric Brownian Motion (GBM) model, which generates the future paths of CPI variables in our simulation.

Homeowners

For owners, we track four distinct indices. Since this data is not available at the metropolitan area or CMA level for the required historical period, we use provincial-level variables. These indices are normalized so that January 2001 equals 1, allowing the simulation to reflect how homeownership-related expenses rise relative to general inflation.

Renters

For renters, the model uses the national CPI for tenant insurance to adjust insurance costs over time, as provincial data for this specific category was not available for the full historical period. Like the owner variables, this index is normalized to 1 in January 2001.

Employment income

Starting salaries are modeled to grow over the 25-year period using Geometric Brownian Motion (GBM) based on historical data. We use provincial wage growth data from Statistics Canada's "Average weekly earnings" (Table 14-10-0223-01). By dynamically increasing incomes, our model ensures realistic future marginal tax rates on capital gains are applied when simulating the end of the scenario.

Applying paths to monthly expenses

For every month in the simulation, the starting parameters established for 2026 are adjusted using the future economic paths generated by our GBM and CIR models. Here is how those starting values are updated:

• Employment income → Average earnings path (Quebec average)
• Rent → Hybrid rent path (Montreal CMHC and Quebec CPI average)
• Renter monthly insurance → CPI Tenant insurance path (Canadian average)
• Property price → CREA Montreal apartment price path (Montreal average)
• Property maintenance cost → CPI Maintenance cost path (Quebec average)
• Condo fees → CPI Maintenance cost path (Quebec average)
• Property tax → CPI Property tax path (Quebec average)
• Buyer monthly insurance → CPI Homeowners insurance path (Quebec average)
• Fixed-sale fees → CPI All-items path (Quebec average)

Monthly expenses

The model calculates the out-of-pocket costs for each scenario. Some expenses only happen once, at the very beginning, while others are recurring every month.

Initial expenses (First month only)

Here are the details on the costs included at the start of the simulation:

Recurring expenses (Every month)

Here are the details on the costs included every month of the simulation:

Monthly gains

This is where the model calculates how much the user's wealth grows each month. Here are the details on what is included:

In the first month of the simulation, the buyer pays a large amount for the down payment and purchase fees. The renter, who avoids these initial costs, is usually able to invest a significant sum into the stock market right at the start.

Balance before selling

This represents the user's total net worth at any given time. It is calculated as the difference between all the wealth the user has generated (the user's assets) and every dollar the user has spent (the user's expenses) since the start of the simulation.

If the balance is positive, it means the user's assets are worth more than what the user has paid out so far, which means the user is "making money." If it is negative, it means the user's total expenses have outweighed the current value of the user's assets. Basically, it is the "bottom line" of the user's financial journey in each scenario.

Here are the details on the assets included in this calculation:

Balance after selling

To truly compare the scenarios on a level playing field, we calculate the "Balance after selling." This represents the "final score" of each iteration by simulating a hypothetical scenario where the user sells their home, cashes out their investments, and pays all outstanding taxes and fees.

This allows us to account for the hidden costs of each choice, such as real estate commissions, mortgage penalties, and capital gains taxes, which are often overlooked in simple comparisons.

Here are the details on what is included:

In real life, people rarely liquidate all their assets at once. For example, selling a large stock portfolio in a single year could trigger a much higher tax bill than selling it gradually over several years.

We use this "total liquidation" approach as a standard way to measure the total tax and fee burden of each scenario. It is not intended as a prediction of how someone would actually manage their exit, but rather as an indication of the final wealth generated after all obligations are met. Finally, since our simulation covers 25 years, the mortgage is fully paid off by the end, meaning no mortgage penalty is actually paid at that point. And since we simulate for a couple, the renters usually have all of their investment in a TFSA, which means they don't pay any capital gains tax at the end of the scenario.

Today's dollars and inflation

Our model provides an option to display all results in "today's dollars." When this option is enabled, all monetary outputs—including assets, expenses, balances, home values, and employment income—are automatically discounted using the specific inflation path generated for that simulation.

This means that if a scenario randomly generates a high inflation rate, the future dollar values are discounted more heavily. This approach ensures that the results reflect the actual purchasing power of the generated wealth, allowing for a more intuitive comparison with present-day costs. All the pre-calculated findings presented in the charts on this page use this inflation-adjustment method.

The first month trade-off

The first month can tell us a lot about how things could end up. The table below compares the initial cash outflow for buyers of a benchmark condo versus renters across different metropolitan areas.

The table helps visualize the trade-offs through four main metrics:

• Price-to-Rent (P/R) Ratio: The purchase price divided by the annual rent. A low ratio generally favors buying, while a high ratio typically favors renting.
• Buy Premium (%): How much more (or less) expensive the buyer's recurring monthly costs are compared to the renter's. A positive percentage means owning requires a higher monthly budget.
• Renter first month savings: The total lump sum that a renter can immediately invest in the stock market because they avoided the buyer's higher upfront and recurring costs in month one.
• Renter win frequency: The percentage of our 25-year simulations where the renter ended up with more wealth than the buyer for that specific metropolitan area and portfolio.

Renting often becomes the stronger financial choice when three factors align:

• High P/R ratio: Buying the property is expensive compared to the cost of renting it.
• High buy premium: Renting is much cheaper on a monthly basis, freeing up cash for alternative investments.
• Large first-month savings: Avoiding the buyer's upfront costs gives renters a significant initial capital boost that compounds over time.

Click on a metropolitan area to see the detailed breakdown.

Mortgage qualification and lending limits

While our simulation compares the long-term wealth of buyers and renters, we also analyzed whether buying is even an option for the average household today.

For this analysis, we asked Statistics Canada for detailed data by deciles for multiple household and individual types, including couple families, one-parent families, individuals not in census families, and four age-based brackets. This approach allows us to see how qualification varies not just by metropolitan area, but by life stage and family composition.

Since the latest available data is for 2023, we projected these income deciles forward to January 2026 to match the start of our simulation. This projection was calculated using the actual provincial wage growth observed between 2023 and December 2025 in Statistics Canada's average weekly earnings data.

Conclusion

As mentioned at the beginning of this page, all the code used for the calculations is open source and available within the journalism TypeScript library (journalism library, simulation logic, monte carlo engine).

For any questions or comments regarding this methodology, please reach out to CBC News Senior Data Producer Nael Shiab.