Measuring The Effect Of A Market Structure Change
When an exchange changes a rule — a new tick size, a new order type, a new fee schedule — proving what the change actually did requires comparing the affected market to one that didn't change, not just watching a before-and-after chart.
Prerequisites: What Makes A Market Good?, Measuring Price Efficiency
An exchange announces a rule change — a wider minimum tick size, a new speed bump, a new fee schedule. Six months later, spreads on the affected stocks are 20% wider than before. Did the rule change cause that, or did the market just get more volatile over those six months for unrelated reasons — a rate hike, an earnings season, a broad selloff? A simple before-and-after comparison can't tell the difference, and getting this wrong is one of the most common mistakes in market structure research.
Why "before versus after" is not enough
Markets move for a thousand reasons that have nothing to do with any specific rule change: overall volatility rises and falls, trading volumes drift with the economic cycle, other unrelated rules change around the same time. If you only compare the affected stocks before the change to the same stocks after the change, you're mixing the effect you care about with everything else that happened to the whole market over that period. A spread that widened by 20% could be 15 points of "the whole market got choppier" and only 5 points of "this rule actually hurt liquidity" — and the raw before-after number can't separate them.
The fix: a control group
The standard solution is to find a group of similar stocks that the rule change did not apply to, and track them over the exact same period. If both the treated stocks and the untreated control stocks got choppier by the same amount, that shared movement is the market-wide noise you want to net out — the true effect of the rule is the difference in how much the treated group changed versus how much the control group changed. This is the core idea behind Difference-In-Differences In Microstructure Research, and it's the standard tool for any "did this policy work" question in market structure.
A worked comparison:
| Group | Average spread, before | Average spread, after | Change |
|---|---|---|---|
| Treated stocks (rule applied) | 4.0 cents | 5.5 cents | +1.5 cents |
| Control stocks (rule not applied) | 3.8 cents | 4.9 cents | +1.1 cents |
Both groups' spreads widened — the market as a whole got more expensive to trade in over this window. But the treated group widened by 0.4 cents more than the control group did. That 0.4-cent gap, not the raw 1.5-cent change, is the researcher's best estimate of the rule's actual effect, because it's the part of the move that the control group — exposed to the same market-wide conditions but not to the rule — didn't share.
The real example this played out in
The SEC's 2016-2018 Tick Size Pilot is the canonical case: a group of small-cap stocks was randomly assigned wider minimum tick sizes (a nickel instead of a penny) while a control group of similar small caps kept the penny tick. Spreads on the pilot stocks widened sharply, exactly as mechanically expected from a larger minimum increment — but trading volume and depth also changed in ways that were only visible by comparing against the untreated control group, not by watching the pilot stocks alone. The Tick Size Pilot And What It Showed covers what the pilot actually found; Decimalisation And Its Effects covers the earlier, opposite-direction change from eighths and sixteenths down to pennies.
A believable claim about what a market structure change did requires a control group that experienced the same broad market conditions but not the rule change itself. The size of the effect is the gap between how the treated and control groups moved, not the raw before-after change in the treated group alone.
If an interview question describes a market structure change and asks you to evaluate its effect, your first move should be "what's the control group here?" — a change that applies to every stock in a market at once (with no untreated control) is much harder to evaluate cleanly than one that was rolled out to some stocks and not others.
Related concepts
Practice in interviews
Further reading
- Harris, Trading and Exchanges (ch. 27)
- SEC, Tick Size Pilot Plan (2016)