OSC-funded study finds Canadian retail pays 32% more in effective spreads than US retail, but no significant gap after toxicity adjustment

The study

The paper is titled “Retail Execution Quality” and dated August 2026. Its authors are Anne Haubo Dyhrberg (University of Delaware), Björn Hagströmer (Stockholm University and Sveriges Riksbank), Andriy Shkilko (Wilfrid Laurier University and University of Georgia) and Ingrid M. Werner (The Ohio State University and CEPR). The OSC’s Research Grant Competition funded the work and has posted it on its website. The paper says its views “do not necessarily reflect the views of the OSC” and that funding or publication “does not constitute endorsement”.

The authors use TMX Group account-level data for the TSX and TSX Alpha from 1 January to 30 June 2023. The data identify retail traders, institutions, designated market makers (DMMs), proprietary traders and high-frequency traders (HFTs). The Canadian sample covers 2,129 TSX-listed securities. The cross-border comparison uses the 146 that are interlisted in the US. US retail execution quality comes from the SEC Rule 605 reports of “the eight largest wholesalers”.

The data are more than three years old. The study measures marketable orders only and excludes commissions and exchange fees. TMX venues account for 54% of Canadian dollar volume in the sample.

Canada against the US

For retail orders in the interlisted stocks, the US quoted spread was 6.80 basis points (bp), against 4.66bp on TMX venues. The ranking reverses for effective spreads. US retail paid 3.91bp and Canadian retail paid 5.00bp. The authors write that price improvement in the US “pushes effective spreads approximately 22% below those paid by Canadian retail investors, despite quotes that are 46% wider.”

After controls, the authors estimate a 1.25bp difference, which “implies a cost premium of 32% for Canadian retail investors.” The differences in realised spreads, their toxicity-adjusted measure, “are not statistically significant in either the retail or the non-retail comparison.”

Non-retail investors in Canada come out ahead. Their quoted spreads are “lower by almost one basis point, or 17%”, their effective spreads are lower by 10% and their price impact is lower by 5%.

A summary that reads differently from the paper

The OSC’s research page summarises the project as finding “that retail execution quality is statistically indistinguishable between the two countries.” The paper itself is narrower. The effective spread difference that favours US retail is statistically significant. Parity holds only on the realised spread. The authors also say that which of the two measures is more appropriate “depends on whether the higher price impact of Canadian retail flow reflects genuinely higher toxicity”. They argue that their Canadian results favour a different explanation: measured price impact is higher because a pooled market cannot easily tell retail orders from other flow.

Inside the Canadian market

• Counterparties. On the TSX, retail marketable orders filled by institutions left those institutions a realised spread of 0.63bp. The equivalent figures were 4.93bp for DMMs and 3.72bp for HFTs.
• TSX Alpha. Retail orders on TSX Alpha paid effective spreads 1.1bp higher than on the TSX. Price impact was 4.0bp lower, which meant realised spreads were 5.1bp higher. During the sample period, Alpha applied a randomised delay to all incoming orders except post-only orders.
• Odd lots. Retail odd lots paid effective spreads 1.2bp wider than board lots. Realised spreads were 5.2bp higher at the ten-second horizon. The counterparties were almost exclusively DMMs, “consistent with the odd-lot facility serving as a supporting revenue stream for DMM activity.”

The authors conclude that the results “point to a trade-off in the design of retail market structure rather than to a ranking of the two systems.”

Why it matters

Analysis. The US debate over wholesaler internalisation usually centres on price improvement measured against the NBBO. This paper takes a different approach: it compares the same companies traded under two market designs. Its finding is that market structure mainly decides who bears the cost of supplying liquidity, rather than whether retail investors as a group are better or worse off. Read that way, both sides of the US argument over order segmentation can cite it. Market practitioners will note the limits: the data cover the first half of 2023, and the study covers marketable orders only.

What we are watching

• Whether the OSC or the Canadian Investment Regulatory Organization responds to the TSX Alpha and odd-lot findings.
• Whether the paper is cited in US comment letters on order competition and Rule 605.
• Whether the OSC revises its summary of the findings.