Estimating the top tail of the family wealth distribution in Canada – 2025 update

Summary

During the 2025 federal election, the Parliamentary Budget Officer (PBO) costed multiple election proposals related to high-net-worth families and wealth taxes. This necessitated an update of the High-net-worth Family Database (HFD) using the most recent data. This report outlines key results and trends from the updated HFD. The methodology used is consistent with previous reports, which was first described in PBO’s June 2020 report titled “Estimating the top tail of the family wealth distribution in Canada.”[^1]

The development of the PBO’s High-net-worth Family Database (HFD) provides a method to address gaps and underreporting in data on high-net-worth individuals which would otherwise distort analysis of wealth distribution. This latest update to the HFD benefitted from the availability of additional data sources compared to previous reports. Specifically:

• The full Survey of Financial Security (SFS) data for the 2023, 2019, and 2016 cycles from Statistics Canada; and,
• Maclean’s 2024 list of Canada’s Richest People.

These sources improved our HFD by providing more recent information about more high-net-worth individuals compared to the SFS Public Use Microdata File (PUMF), the discontinued Canadian Business rich list last published in 2017, and the Forbes billionaire list used in the previous reports.

This report presents the updated distribution of family net wealth for 2023 using the updated HFD and compares it to the 2023 SFS. This report also examines the change in this distribution from 2016 to 2019 and 2023, and discusses how different data sources and sampling uncertainty may affect the interpretation of these results.

The updated HFD suggests that by comparison, the concentration of family net wealth at the top end, particularly in the top one per cent, is considerably understated by the 2023 SFS (Table ES-1). The HFD indicates the top one per cent of economic families own around 23.8 per cent of net wealth in Canada versus the 11.0 per cent suggested by the 2023 SFS. The top 20 per cent of economic families own around 69.2 per cent of net wealth in Canada (vs. 63.5 per cent suggested by the 2023 SFS), while the middle and bottom 40 per cent own around 27.6 and 3.3 per cent of net wealth (vs. 32.6 per cent and 3.9 per cent suggested by the 2023 SFS).

When comparing wealth concentration from 2016 to 2023, PBO’s HFD confirms the slight increase from 2016 to 2019 observed in the previous report but notes a reversal of this trend from 2019 to 2023 (Figure ES-1). Given the uncertainties in the data, the concentration of wealth among high-net-worth families in 2023 is considered to be either comparable to or marginally lower than in 2019.

PBO’S HFD and Statistics Canada’s SFS (2016, 2019, and 2023).

PBO’S HFD and Statistics Canada’s SFS (2016, 2019, and 2023).

PBO also compared the differences of using the PUMF vs. the full SFS data for constructing the HFD and created a public copy of the PBO’s 2023 High-net-worth Families Database, which is available on the PBO’s website.[^3]

Introduction

An initiative born from a lack of publicly available data that reliably assesses the top end of family wealth distribution in Canada, the Parliamentary Budget Officer (PBO) created the High-net-worth Families Database (HFD). The HFD offers an approach to address underreported and missing data of high-net-worth individuals that would otherwise distort analysis of wealth distribution.

The initial creation of PBO’s HFD in 2020 greatly improved PBO’s capacity to undertake quality analysis of parliamentary initiatives affecting high-net-worth individuals. The subsequent update to the HFD expanded on this capacity, allowing for temporal analysis on wealth distribution which was published in our 2021 report.[^4]

Using the same modelling approach as previous publications, this latest HFD update offers access to more detailed and recent data, improving the quality and accuracy of the results. Generally, a Pareto interpolation was used to bridge the gap between the latest SFS microdata (2023) and publicly available billionaire lists (Forbes and Maclean’s) to create a more representative synthetic dataset for high-net-worth families compared to the unadjusted SFS. This dataset was then adjusted to match the aggregate household wealth according to the National Balance Sheet Accounts (NBSA).

This report proceeds as follows. First, a description of how the HFD was updated to reflect the latest data from Statistics Canada’s 2023 SFS, as well as from the corresponding billionaire lists from Forbes and Maclean’s. Then, the report compares wealth distribution trends from 2016 to 2023 using consistent data, including a discussion on details of net worth by asset/debt categories as well as the impacts of sampling uncertainty. Finally, the report summarizes the sources of uncertainty and limitations.

Estimating the 2023 HFD

Data and Methodology

This report largely follows Vermeulen’s (2018) methodology, as in previous PBO reports, using updated data.

The new data includes the 2023 cycle of the Survey of Financial Security (SFS) and updated rich lists from Forbes[^5] and Maclean’s Canada’s Richest People.

For the SFS, the PBO accessed the full microdata file from Statistics Canada’s Research Data Centre, as opposed to the Public Use Microdata Files (PUMFs) used in the previous reports.

For the rich lists, PBO used updated information from Forbes and Maclean’s 2024 publication Canada’s Richest People.[^6] Adjustments were made to align the timeframes of these two data sources.

The SFS does not provide a reliable sample of families with a very high net worth.[^7] To correct for this, PBO used Pareto interpolation to estimate the net worths of families between the adjusted billionaire lists and the upper ranges of the SFS. The additional families generated through this interpolation were added to the overall database, and the total net worth was adjusted to align with Canada’s total household net worth according to Statistics Canada’s NBSA.

For the final HFD 2023, PBO used the Forbes data source supplemented with Maclean’s when the latter contained additional billionaires not on the former list. However, an alternate version of the HFD using only Forbes was also estimated, especially for comparison with past years when only the Forbes list was available. The final HFD dataset was then projected to the last quarter of 2024 by aligning it with more recent historical data on population and NBSA household wealth in Canada.[^8]

Results

The 2023 HFD shows that the top one per cent of high-net-worth families in Canada hold 23.8 per cent of the country’s total net wealth. Like PBO’s previous reports, the 2023 HFD showed significantly higher wealth concentration (by 12.8 percentage points for the top one percent) compared to the unadjusted SFS data. The difference between SFS and HFD becomes much narrower below the 80th percentile of economic families by net worth. Table 1 shows the share of total net wealth by selected percentiles.

Table 2 displays more details on the selected percentiles from the HFD. It summarizes the net wealth thresholds of each selected percentile, along with the number of families, total net wealth, and share of overall net wealth within the selected percentile range. The threshold for top one per cent of family net wealth is $7.4 million in 2023.

Table 3 presents additional net wealth summary statistics by wealth thresholds instead of percentiles. In 2023, there were 4.4 million families in Canada with net wealth above one million dollars and 108,000 families with net wealth above 10 million dollars. The updated HFD captures a more complete list of billionaires, suggesting a higher concentration in wealth among the highest absolute thresholds (i.e. over $1 billion) than previously reported in our 2019 report.

The 2023 HFD was also used as a baseline to update the database to Q4 2024 using available NBSA, rich list, and population data. The updated Q4 2024 database was used as the baseline for costings related to wealth taxes during the 2025 federal election. The methodology and assumptions are consistent with PBO’s 2020 and 2021 reports. Table 4 summarizes the net wealth distribution by selected percentiles in Q4 2024.

Evolution of Wealth Concentration, 2016 - 2023

In this section, we provide an update to the evolution of wealth distribution from 2016 to 2023.[^10]

Figures 1 and 2 compare the share of wealth of top 1 per cent and 5 per cent respectively from 2016 to 2023 using three data sources: the SFS data only, the HFD constructed using SFS and the Forbes billionaires list, and the HFD constructed using SFS, Forbes, and the Canadian Business or Maclean’s billionaires list.

PBO’S HFD and Statistics Canada’s SFS (2016, 2019, and 2023).

PBO’S HFD and Statistics Canada’s SFS (2016, 2019, and 2023).

PBO’S HFD and Statistics Canada’s SFS (2016, 2019, and 2023).

PBO’S HFD and Statistics Canada’s SFS (2016, 2019, and 2023).

Overall, there was an increase in the percentage of wealth held by the top one per cent of economic families from 2016 to 2019, followed by a decrease from 2019 to 2023. In the SFS and SFS + Forbes series, the top one per cent held less wealth in 2023 compared to 2019, and in the SFS + Forbes + CB/Maclean’s series (where 2019 was not available), the top one per cent held less wealth in 2023 compared to 2016. The share of wealth held by the top 5 per cent largely followed the same trends.

Caution must be exercised when interpreting this result. The Forbes dataset does not contain a consistent list of individuals in all years: for example, some individuals on the 2019 Forbes list were not present on the 2023 or 2024 list, whereas they were captured by the Maclean’s list (or vice versa). This means that the 2023 estimate derived from Forbes may underestimate the true wealth concentration. In addition, the SFS does not have sufficient sample sizes to estimate the net worth of the top one per cent of the distribution, and there is significant sampling uncertainty involved. Therefore, direct comparison of the point estimates in different years may be misleading.

However, the SFS is designed to create representative estimates of net worth up to the top 5 per cent of the distribution.[^11] Figure 3 illustrates the sampling uncertainty by showing the 95 per cent confidence intervals for the share of net wealth held by the top 5 per cent according to the SFS, providing a visual sense of the variability in the estimates.

Since the SFS is a sample survey, it is important to consider sampling uncertainty and statistical significance when interpreting the changes in wealth concentration over time.

Statistical significance is a concept for determining whether there is evidence to conclude a result of a study is too unlikely to have occurred by chance. In this case, we are studying whether the difference in the percentage of wealth held by the top 5 per cent of households in two different years. We start with a null hypothesis (i.e. default assumption) that there is no difference in the percentage of wealth held. Then we observe the survey results and determine the probability of observing a difference at least as large as what the surveys reported under the assumption that the null hypothesis is true. If this probability is under 5 per cent, we say that the difference in wealth percentage is statistically significant.

SFS provided a set of 1,000 bootstrap weights for constructing confidence intervals of percentile results. To test whether the difference between two years is statistically significant, we calculate the differences across 1,000 pairs of bootstrap samples and construct a 95 per cent confidence interval. If zero is not contained within this interval, then the difference is statistically significant (and vice versa).

Statistics Canada’s SFS (2016, 2019, and 2023).

Statistics Canada’s SFS (2016, 2019, and 2023).

In particular, the SFS data itself showed a large decline in the percentage of wealth held by the top 5 per cent from 2019 to 2023 (by about 4 percentage points), following almost no change from 2016 to 2019. Moreover, this change was statistically significant (further discussed in Appendix B).[^12] However, it can be noted from Figure 2 that the gap in wealth concentration from 2019 to 2023 became much smaller after combining the SFS with the Forbes rich list (by around 2 percentage points). The difference was still statistically significant.

Details of net worth by asset/debt category

Due to methodological updates, the 2023 HFD does not contain details on the categories of net wealth (that is, assets and debt). Looking at the SFS offers insights into the breakdown and trend of these categories as a percentage of total net worth held by the 5 per cent wealthiest economic families (Table 5).

Based on Table 5, financial assets without pensions, non-financial assets, and total debt held by the top 5 per cent remained a roughly constant share of total net worth from 2016 – 2023, but the share of pension assets and business equity dropped from 2019 to 2023, contributing to an overall drop in net worth held by this group. However, business equity values in the SFS had much higher imputation rates in 2023 than in 2019, which may affect the reliability of comparisons over time.[^13]

Sources of Uncertainty

There are some caveats and limitations related to the creation of the HFD and the interpretation of wealth concentration changes over time. The different data sources of the billionaire lists (Forbes, Canadian Business, and Maclean’s) include different sets of individuals, and even the same Forbes list include different individuals in different years. The Pareto estimation is sensitive to both the composition of the billionaire list and the reported net worth figures, so uncertainty at the top end can affect estimates across the entire HFD. In addition, net worth can fluctuate significantly within a year, adding uncertainty when aligning data from sources with different reporting frequencies. Statistics Canada’s SFS data also carries uncertainty, stemming from both sampling variability and differing imputation rates across survey cycles, which can complicate direct comparisons of wealth concentration over time.

Difference between PUMF and full SFS data

One major data improvement of this report compared to the previous reports in 2020 and 2021 is the use of full SFS data instead of the PUMF data used previously. There are slight differences between the PUMF and full datasets to preserve the confidentiality of respondents in the public microdata. However, this could introduce an additional distortion when estimating wealth distribution across the population. Table A-1 compares the unadjusted wealth distribution from Statistics Canada’s SFS 2023 PUMF and full data, as well as the final PBO’S HFD 2023 estimated using the SFS PUMF and the full SFS.

For both 2023 and 2019, there was a small difference between the PUMF and full SFS data (Table A-2). The proportions of wealth held by each percentile group were almost always within one percentage point between the two data sources. When the billionaire list was added to create the full HFD, this difference remained or sometimes became larger. The full HFD’s wealth proportions held by each percentile group was usually around 1 to 3 percentage points lower than the PUMF’s corresponding proportion. This is because the difference between the two sources is largely among the top records, which leads to an amplified impact on the estimation of the Pareto parameter and thus the interpolation between the SFS top records and the billionaire list.

Therefore, while having access to the full SFS data can provide a slightly more accurate estimate of wealth distribution, the differences are usually not large compared to the PUMF data.

Sampling Uncertainty

Sampling Uncertainty in the SFS

As shown in Figure 3 previously, the confidence intervals for the share of wealth held by the top 5 per cent did not overlap in 2019 and 2023. While checking for overlap in confidence intervals is not the exact test for statistical significance, it can provide a reasonable quick approximation.

To confirm whether the difference from 2019 to 2023 is statistically significant, we used the bootstrap samples provided by Statistics Canada alongside the SFS data to construct a distribution of sample differences across the two years. The middle 95 per cent of these bootstrap sample differences formed the 95 per cent confidence interval. It was found that zero fell outside the confidence interval for the difference between 2019 and 2023, indicating that the drop from 2019 to 2023 was statistically significant.

The hypothesis test showed that there was evidence to reject the null hypothesis that there was no actual difference in wealth concentration among the top 5 per cent of economic families in 2019 and 2023 and that the observed difference was purely due to the randomness of the chosen sample. In other words, there is evidence that the observed drop in wealth concentration (about four percentage points from 2019 to 2023) is too large to be simply attributed to chance.

Of the different components of net worth, pension assets showed a drop of three percentage points from 2019 to 2023, and this difference was statistically significant. In addition, the share of debt was also about 0.4 percentage points higher in 2023 than in 2019, with the difference being statistically significant. The other components did not show a statistically significant difference between 2019 and 2023. Table B-1 shows the 95 per cent confidence intervals of the change in each asset/debt component in consecutive SFS cycles.

Sampling Uncertainty in SFS + Forbes

The methodology for calculating confidence intervals for the combined SFS + Forbes data was more complex. The bootstrap samples were also used to determine the confidence intervals for the share of wealth held by the top 5 per cent according to the SFS + Forbes data (see Figure 2). A Pareto interpolation of the top tail was estimated using each SFS bootstrap sample alongside the Forbes data to create a “bootstrap HFD”, and the percentage of net wealth held by the top 5 per cent was calculated for each “bootstrap HFD” data year. The middle 95 per cent of these bootstrap sample differences formed the 95 per cent confidence interval. In this case, zero fell outside the 95 per cent confidence interval for the difference between 2019 and 2023, indicating that even after combining SFS with Forbes, the observed drop in wealth concentration between 2019 and 2023 was still statistically significant. Table B-2 reports these confidence intervals.

References

• Canadian Business (2016). “Canada’s Richest People: The Complete Top 100 Ranking” (2017).

• Forbes. (n.d.). “The World’s Real-time Billionaires”.

• Maclean’s. (2024). “Canada’s Richest People”.

• Office of the Parliamentary Budget Officer. (2020). Estimating the top tail of the family wealth distribution in Canada.

• Office of the Parliamentary Budget Officer. (2021). Estimating the top tail of the family wealth distribution in Canada: updates and trends.

• Vermeulen, P. (2018). “How Fat is the Top Tail of the Wealth Distribution?”. Review of Income and Wealth, 64(2), 357-387.

• Statistics Canada. (2024, October 28). Survey of Financial Security (SFS).