Why Bid-Ask Spreads Contribute to Market Impact
Summary
The document explains why some price-impact models include a term proportional to the bid-ask spread. When execution must be assured, a trader may have to cross the spread and take liquidity; the model coefficient can represent the chance that this becomes necessary. In midpoint auctions or dark pools, order imbalances may move prices away from the midpoint, reducing the apparent benefit of midpoint execution.
A further channel is dealer hedging: a market maker who takes the other side may need to offset inventory by trading immediately and crossing the spread, with that cost reflected in quotes. These mechanisms explain why a narrower spread can correspond to lower execution costs and impact. The answer does not estimate the model coefficient or establish a universal linear relationship; it notes that multiple impact models exist and that impact can be temporary, decaying, or permanent. The discussion is qualitative and does not provide empirical evidence or a calibration method.
Key ideas
- Urgent execution may require crossing the spread, and the model coefficient can capture the likelihood of doing so.
- Order imbalances in midpoint matching venues can shift prices away from the midpoint.
- Market makers may pass spread costs from hedging their inventory into their quotes.
- The spread term offers several possible mechanisms, but the relationship depends on the impact model and trading context.
- Price impact may be temporary, decaying, or permanent.
Tags
Full text
# Market Impact proportional to the bid-ask spread
# Market Impact proportional to the bid-ask spread
Empirical studies have shown that market impact can be linked to the following parameters:
$$ \mathcal{I}(Q) = \kappa \ . \ \sigma \ . \ (\frac{Q}{V})^\gamma + \alpha \ . \ \psi_{BA} $$
where $\psi_{BA}$ is the bid-ask spread.
Can anyone explain the linear relationship in the spread? Why having a narrower spread implies lesser impact?
Thanks!
## Answer by kurtosis (score 3)
https://quant.stackexchange.com/a/55946
There are multiple models for price impact and the one you have listed here is not the latest. You can see a writeup of a few of the most popular and recent models in this answer.
We can think of a few reasons why price impact is considered linear in the bid-ask spread.
First, you want the trade to be completed (implied by most of these models). You may be unable to wait for your price to be taken; you will need to switch from a price maker to a price taker, cross the spread, and trade at the far side fo the market. In that case, you pay the bid-ask spread to guarantee execution. However, that is not certain to occur, so $\alpha$ accounts for that probability.
Second, suppose you trade in dark pools (or other ATSs using occasional auctions/matching at the midpoint). The imbalance of orders in a matching auction will bleed back out to the market and so instead of trading at mid-market when you entered your order, the price will shift a little. In that case, you again end up paying a price that is half the spread plus or minus some shift in that midpoint.
Third, market makers who take on a position when you trade with them eventually need to hedge. Doing that may require them to immediately get out of a position, so they pay the bid-ask spread to trade immediately. Their quotes will therefore reflect the spread to pass that cost on to you.
If you have not traded much, you might want to try doing so in a good simulator or (far better) with some real money. You will quickly see how often you need to consider crossing the spread and why that affects the price you pay.
You should also probably read up on permanent versus temporary (and decaying) price impact at the link above. That may help clarify the thinking of how the bid-ask spread affects prices.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.