11 reasons why government’s claim about money laundering in B.C. housing is false and misleading

This is the story of what happens when poor data and bad math meet a frustrated society in crisis, governed by opportunistic politicians, with journalism nowhere to be found*.

Earlier this month, the B.C. government published two studies that looked at new datasets of real estate transactions. They reported that “more than $7 billion in dirty money was laundered in B.C. in 2018, hiking the cost of buying a home by about 5%.”

Cue in an onslaught of online hysteria, press hearings, indignant reporters asking tough questions from officials caught off-guard by the scandal, even some reporting further inflating the housing price increase to “upwards of 20 percent.”

The problem is, these conclusions were false! They were not based on B.C. real estate data, and were founded on questionable assumptions and bad math. Here’s why:

[*UPDATE: Turns out Bloomberg read the reports and questioned the validity of the claims: “Vancouver’s Dirty Money Figures: The Smoking Gun That Wasn’t”]

1) Contrary to initial reports, these claims were NOT based on B.C. real estate data.

B.C. government published two studies simultaneously. While the media highlighted former RCMP deputy Peter German’s study of B.C. real estate data, they failed to report that German in fact did not find a reliable way to estimate how much money is laundered in B.C.

Instead, they echoed conclusions of a second ”panel of experts” study that used a macroeconomics equation, using parameters such as GDP and crime rates, to make predictions (as opposed to take measurements) of how much money is laundered, while relying on a number of unsubstantiated assumptions.

2) The study used investment portfolios of Canadians to predict how criminals launder money. Big assumption, no evidence.

After estimating B.C. money laundering at $7.4B, the study tried to calculate how much of that money flowed into real estate investments. They used “portfolio allocation decisions of the general public as a benchmark” for how criminals invest dirty money (page 50)!

This is a completely unsubstantiated assumption for which the panel offers no supporting evidence, nor do they attempt to validate their results.

On what grounds is criminal money laundering behavior modeled after Canadians investment behavior? Does this mean criminals also invest in RRSPs and index funds? Or do average Canadians rely on well-known money laundering schemes such as casinos and gambling, cash-intensive businesses, etc. at a similar rate as criminals do?

3) The study assumed that criminals use 100 percent of their dirty money to invest, and live off of the proceeds. Another big assumption without any evidence.

The study’s assumption that criminal “inflows are fully dedicated to investment purposes and domestic criminal consumption is funded by investment returns” (page 52) was backed by no supporting evidence, and frankly sounds hard to believe.

This assumption also contradicts the panel’s earlier assumption that Canadian investment portfolios are a suitable benchmark for money laundering behavior. Average Canadians invest 3.6 percent of their disposable income. High-income Canadians invest 28 percent. These are both far from the 100 percent number used by the study.

4) The study assumed 72 percent of laundered money in B.C. goes into real estate, once again without evidence.

While average Canadians allocate 37 to 72 percent of their investment budgets to real estate (40.6 percent on average), the study picked the highest end of that range for criminal’s real estate allocation. No factual basis for this assumption was presented. This yielded to the much-reported conclusion that $5.3B in B.C. goes into real estate (72 percent of $7.4B).

To put 72 percent in perspective, average Canadians invest no more than 1.5 percent of their earnings into real estate (a 40.6 percent allocation of their 3.6 percent investment budget), and that increases to no more than 20 percent for high-income Canadians (a 72 percent allocation of 28 percent investment budget). That means rather than $5.3B, the correct estimate of dirty money in B.C. real estate could be $1.5B or as low as $0.1B.

5) The study shows a lot of cherry picking, an error born from bias where researchers only use data confirming their thesis, and ignore everything else.

There is a name for when researchers pick and choose numbers in support of their thesis: cherry picking. The panel’s work reads like a case study of the cherry picking fallacy. Besides arbitrarily choosing favorable statistics, the panel does not test for disconfirming evidence or try to disprove its conclusions.

To compare, Peter German’s study tries to invalidate its conclusions frequently and ultimately states that its work may not be a reliable indicator of money laundering in real estate.

The panel’s study and its choices of assumptions smell of bias, particularly since every unsupported assumption by the authors led to increase their estimates.

6) RCMP’s estimate from 2001 was less than one third.

The panel estimated $47B of money laundering in Canada in 2018 (page 47), while mentioning a 2001 RCMP estimate of $5B to $15B. By its nature, money laundering is hard to measure. Therefore it is important to compare any research to other works, particularly for this study given the numerous problems discussed here.

7) B.C. actually ranked fourth in Canada in the study, yet no other region in Canada has supported these findings.

Even with the results that seem inflated, B.C.’s $7.4B estimate ranked fourth behind Ontario, Alberta and the Prairies (Saskatchewan and Manitoba). That means B.C. has the lowest rate of money laundering west of Quebec, according to the study. Looking at it per capita, B.C. ranked worse than the Atlantic region and had almost half the per capita laundering rate of Alberta, Saskatchewan and Manitoba!

Given so many regions with similar or worse problems, how do we explain that the panel’s study stands as an outlier in its evaluation of the impact to housing markets?

One explanation is that even though this problem is more pronounced in other regions, the panel was the first to notice. A more plausible explanation, however, is that given the plethora of unsubstantiated assumptions in this work, the panel’s estimates are off the mark and exaggerated.

8) The study does not adequately acknowledge and disclose its high margin of error.

Based on the panel’s study, the actual share of dirty money in B.C. real estate could be $1.5B or even as low as $0.1B (see #4). In fact the report briefly mentioned a range of $0.8B to $1.5B as a possible alternative to their $5.3B estimate (page 52). But they never again repeated those numbers, including in the conclusions that were summarized on page one.

In its conclusions, the panel asserted its claim of 5 percent housing price impact, while repeatedly calling its estimates “conservative” and “cautious.” Using the same calculations, the housing impact could be as low as 0.1 percent. Would there be the same public reaction if the panel stated that the housing price increase in B.C. could be as low as 0.1 percent?

The panel passingly acknowledged the difficulty of measuring money laundering in its conclusions. After the release of the report and in a press conference, the study’s main author even admitted that the panel questioned whether to publish their findings given concerns about reliability. Nonetheless, those concerns were never adequately highlighted in the report’s conclusions. In particular, it is surprising and disappointing that the panel failed to provide the margin of error of its estimations, a common way in the scientific community to assess and acknowledge how reliable one’s estimates are.

9) If applied, many of the panel recommendations wouldn’t change its estimates. This once again suggests that the estimates are not reliable.

The panel makes 29 policy recommendations to combat money laundering and its impact on real estate. The problem is that if the panel performs its study once again after many of the recommendations are implemented, their estimates won’t improve! That is because the panel’s estimates were not based on actual real estate data in B.C.

It should be noted that some of the panel’s recommendations have raised eyebrows from civil rights organizations. For example, the proposal to allow the government to confiscate properties without finding a crime — the unexplained wealth orders (UWOs) — is one that B.C. Civil Liberties Association called “an incredibly troubling notion.”

10) The two studies were conflated, lending credibility to the wrong conclusions.

Peter German’s study performed some groundbreaking work by looking at new datasets of B.C. real estate transactions. The government chose to publish that alongside the panel’s study, which guised the panel’s conclusions with German’s more scientific analysis, conflating the two reports and lending undue credibility to the panel’s conclusions.

Unfortunately the media simply echoed the government talking points without investigating them.

11) In fact, the studies cast doubt on whether it is possible to make the conclusions the government is making.

Peter German’s study stated this clearly, while the panel only acknowledged it after publishing its report.

After looking at various indicators of money laundering such as using a trust to hide property ownership, Peter German tested his methods against 154 suspected properties. He concluded that these “may not be an effective way to detect properties linked to money laundering” (page 111). As a result, he never attempted to stake a claim on how much money is being laundered in B.C. real estate.

The panel, too, understood that its results may not be reliable at a provincial level, and questioned whether to publish them at all. Unfortunately, however, they failed to acknowledge this within the 184 pages of their report.

Money laundering is bad. Why challenge this report?

Money laundering is indeed a real problem, and it needs our attention. So is the lack of housing affordability in B.C. — a serious problem in need of a serious solution.

But solutions that are based on bad data will fail to solve the underlying problems. The attention and resources wasted on wrong solutions take away from our ability to find and invest in real and effective ones.

Worse, bad solutions have unintended consequences, from giving legitimacy to xenophobic voices in society that are getting bolder every day, to resulting in policies that erode civil rights.

Disclaimer: I am not an economist, an investigator, or a subject matter expert in money laundering. I have a Ph.D. in engineering, with the mathematical training to understand and evaluate data analysis like these. My day job is making robots, but my curiosity about the recent news led me to read the 550 combined pages of both studies. Learning what I learned, and watching the news unfold, I decided to write this piece to ensure policies are not informed by bad data.