The Data Toll
Can an exchange price its rival out of view?
Katya Malinova
McMaster University
Andreas Park
University of Toronto
Munich · 18 September 2026
The research programme
Two papers on barriers to venue competition
THIS PAPER
Separate data fees
Make the rival’s data harder to buy
Squeeze its liquidity
SECOND PAPER
Integrated broker–exchanges
Crypto-style platforms
Explicitly fragmented liquidity
What changes when the broker
is the execution venue?
Today: the data toll
Motivation
Why data fees matter
Motivation
Data matter and the Cost of Data is Controversial
“There’s No Market in Market Data”
Market Structure Partners / AFME · 2025
“Exchanges hit back at ‘inaccurate’ and
‘misleading’ accusations around market data costs”
The TRADE · 2025
Who pays for seeing the market?
Motivation
Brokers pay, exchanges earn
authors' calculations from TMX annual reports
TSX: market-data income relative to trading-fee income
2006 = last year of trading monopoly
| 2006 | 2023 | |
|---|---|---|
| Trading (in $M) | 121 | 45 |
| Subscription data ($M) | 85 | 100 |
| Professional Subscribers (K) | 139 | 101 |
| Data per subscriber ($) | 612 | 990 |
| Trading per subscriber ($) | 871 | 446 |
| Mils per trade ($.0001) | 2 | 4 |
| Data/trading | 70% | 222% |
Motivation
Regulators are rethinking access
Canada
CSA consultations and industry committees
Data fees, retail access, common terms
CSA Consultation Paper 21-403;
CSA Staff Notice 21-334
Europe
ESMA authorises EuroCTP
One consolidated view of shares and ETFs
United States
SEC market-data infrastructure reform
Broader consolidated information
SEC Release 34-90610
Motivation
Competition needs visibility
A challenger's data is worth buying only if others buy it too: they post the orders that make its book worth seeing.
Motivation
This paper: data fees as a competitive weapon
This paper
Three results
1
Challenger data need critical mass
Zero adoption can persist because the data are valuable only when others subscribe.
2
The data toll
The dominant venue can use its own data charge to squeeze or exclude the challenger.
3
Traders lose; brokers can amplify the problem
Lost adoption reduces trader surplus; broker incentives and fixed fees impede access.
The model
Key Model Ingredients
The model
Two venues, a continuum of traders
dominant venue \(d\): everyone has its data
challenger \(c\): data costs \(d_c\)
private value \(v\sim U[0,1]\), trade \(q\) at cost \(c\)
volume \(q^*=v-c\), surplus \(S=(v-c)^2/2\); traders with \(v>c_0\) are active: mass \(\kappa=1-c_0\)
The critical component is the cost \(c\):
- it determines participation in the market
- it determines trading volume and surplus
- it determines whether a trader subscribes to the challenger's data
\[\max_{q\geq0}\;\underbrace{vq}_{\text{trading benefit}}-\underbrace{cq}_{\text{trading cost}}-\underbrace{q^2/2}_{\text{diminishing returns}}\]
The model
Seeing the challenger's book lowers trading costs
cost with challenger data, when a share \(\alpha\) of traders subscribe:
\[ c_0 \;\longrightarrow\; \underbrace{c_0-\delta\alpha}_{\text{more subscribers, deeper challenger book, lower cost}} \]
your data is worth more when others buy it too
high-value traders: \(v>v^*\) subscribe, so \(\alpha=1-v^*\)
α = share buying challenger data
δ = strength of the liquidity benefit
\(d_c\)=challenger fee ; dominant venue's data fee comes later
Higher-value traders trade more, so they gain more from the challenger data.
Result 1 · Coordination
The challenger's subscription rate is fragile
Result 1 · Coordination
Buying data is a coordination problem
I will buy the feed…
…if enough others buy it too.
A higher fee raises the number of other users needed to make subscribing worthwhile.
Result 1 · Coordination
The same equilibria as a fixed-point problem
Actual
adoption
Expected adoption
Above the diagonal:
adoption grows
Below the diagonal:
adoption falls
Same posted challenger fee. A higher metered toll shifts the blue response down.
Result 1 · Coordination
The fee the marginal subscriber will pay, given that a share \(\alpha\) of others subscribe
\[ d_c \;=\; \underbrace{\delta\kappa\alpha}_{\text{value rises with adoption}} \;-\; \underbrace{\tfrac{\delta(2-\delta)}{2}\,\alpha^2}_{\text{marginal subscriber has lower } v} \]
the maximum \(\bar d_c=\dfrac{\delta\kappa^2}{2(2-\delta)}\) is the viability bound: above it, nobody buys
Result 1 · Coordination
A challenger needs critical mass
Challenger-data adoption
Sustainable
data fee
0
Posted fee \(d_c\)
No adoption
Critical mass
High adoption
unravels
adoption grows
Dashed line: the same subscription price for every buyer.
Result 1 · Coordination
What moves the hump
higher \(\delta\) or \(\kappa\): raise the viability bound, lower the critical mass, raise \(\alpha_H\)
higher \(d_c\): raises the critical mass, lowers \(\alpha_H\)
total volume \(= \underbrace{\kappa^2/2}_{\text{no challenger data}} + \underbrace{\delta\alpha^2}_{\text{gain from adoption}}\)
coordination failure (everyone stuck at zero) costs the market \(\delta\alpha_H^2\) in volume
Result 2 · The data toll
The data toll
Result 2 · The data toll
The Dominant Venue also charges for data
Fixed subscription
Pay once per period
\[d_d\]
Same payment with one feed or two
Metered usage
Pay more when using more
\[m\times q\]
Raises the cost of another trade
Participation
Participation + trading intensity
fixed subscription \(d_d\) and a metered charge \(m\) per unit traded; both exist in practice ("metered is "per quote" so "per trade" is an approximation)
Result 2 · The data toll
Can own data fees weaken demand for rival data?
Raise the price of the dominant venue’s data…
Metered charge
Raises marginal trading cost
\(\kappa(m)=\kappa_0-m\)
Lower viability; higher critical mass
Fixed subscription
Cancels from the relative data choice
for traders active either way
Works through participation
Result 2 · The data toll
Why a metered charge weakens the challenger
\[\underbrace{\text{saving per trade}}_{\delta\alpha}\;\times\;\underbrace{\text{trading activity}}_{\text{falls when }m\text{ rises}}\]
Fewer trades to save money on
Less willingness to pay for the challenger data
Intuition for the incremental surplus from subscribing.
Result 2 · The squeeze
A metered fee squeezes the challenger
Challenger-data adoption
Sustainable
data fee
0
Posted fee \(d_c\)
No metered fee
Higher metered fee
Critical mass rises
High adoption falls
Dashed line: the challenger’s posted subscription fee stays unchanged.
Result 2 · Profitability
Does the data toll pay for the dominant venue?
Result 2 · Profitability
When does squeezing pay?
Lost trading activity
Recaptured order flow
The second effect can outweigh the first
Result 2 · Order-flow recapture
Traders who drop challenger data reroute their orders
Buy challenger data
Drop challenger data
Subscribers route more flow to the challenger than non-subscribers do: σ > λ.
Result 2 · Local incentive
When does a further fee increase pay?
Fixed fee
A further increase lowers total profit
Metered fee
A further increase can raise total profit
when active non-subscribers remain
and the challenger had captured enough subscriber flow
Why: higher \(m\) lowers adoption; volume moves from subscribers (route \(\sigma\) to \(c\)) to non-subscribers (route \(\lambda<\sigma\)).
\(\hat\sigma\) is low when \(\lambda\) and \(\delta\) are small, and falls as the challenger nears its viability bound.
The toll pays when the challenger is taking sufficient trading income from the dominant venue's subscribers.
Reserve · Local profitability
The local profitability threshold
\[\sigma>\widehat\sigma=\lambda+\frac{1-\lambda}{1+\delta}\left[\frac{\kappa_M[(2-\delta)\alpha_H-\kappa_M]}{\alpha_H^2}+2\delta\right]\]
σ: challenger share of subscriber flow
λ: challenger share of non-subscriber flow
Closer to the viability bound makes the threshold lower
At an interior metered data-revenue optimum; active non-subscribers remain.
Result 2 · Global exclusion
A larger fee can eliminate adoption
Challenger-data adoption
Sustainable
data fee
0
Posted fee \(d_c\)
No metered fee
Higher metered fee
No positive intersection
The challenger’s fee has not changed
Result 2 · Global exclusion
Either fee can exclude a fragile challenger
Different channels — the same loss of challenger-data adoption
Fixed subscription
Fewer traders participate
↓
Less support for challenger adoption
Metered charge
Trading becomes more costly
↓
Less willingness to buy challenger data
Near the challenger’s viability limit,
recaptured trading revenue can make exclusion the best fee choice
Given challenger offer; under the paper’s trading-fee and routing conditions.
Result 2 · Global exclusion
A fragile challenger can be worth excluding
Near its viability limit
Only a small squeeze is needed to remove positive adoption
Either tariff can make exclusion the best response
Fixed fees work through participation. Metered fees directly lower the benefit of the challenger data.
Result 3a · Trader surplus
The challenger benefits traders
Result 3a · Trader surplus
Traders lose from lost visibility
At a given dominant-venue tariff
\[\underbrace{W(\alpha_H)-W(0)}_{\text{trader surplus gained}}\;=\;\frac{\delta\alpha_H^3}{2}\;>\;0\]
Losing adoption removes better trading opportunities
Closed form when subscribers would trade anyway. The positive ranking holds more generally.
Result 3b · Broker incentives
Brokers can amplify the problem
Result 3b · Broker incentives
Brokers are gatekeepers
The broker pays the bill
The client gets the full benefit
Result 3b · Broker incentives
Agency acts like a data-price markup
\[\underbrace{\beta}_{\text{share internalised}}\times\text{client benefit}\;>\;\text{data bill}\]
Internalise half the benefit…
…act as if data cost twice as much
With a severe enough agency wedge, positive adoption disappears.
Result 3b · Broker incentives
Fixed distribution fees favour large brokers
More clients over whom to spread the cost
fixed per-firm fee \(F_c\) \(\to\) cutoff \(n^*=\bar n\sqrt{1-\alpha}\):
Takeaways
Takeaways
Takeaways
Three results — and why fee design matters
1
Liquidity needs traders who can see it
Challenger data need a critical mass of subscribers.
2
The Dominant Platform may be able to squeeze the challenger by raising its fees
Locally, metered fees can pay by squeezing adoption; fixed fees do not.
Globally, either fee can make exclusion profitable through different channels.
3
Traders lose; brokers can amplify the problem
Lost adoption reduces surplus. Agency and fixed distribution fees impede access.
Local comparison: at the fee that maximises data revenue.
Takeaways
Does a consolidated tape fix it?
a voluntary tape whose new content is the challenger's data faces the same adoption problem
the tape's extra cost plays the role of \(d_c\)
professional adoption may seed challenger liquidity, but nothing guarantees it
no requirement that brokers show the tape to clients
Supplying a tape does not by itself ensure adoption or display.
Policy implications
A consolidated tape still needs to reach traders
Data available
Data displayed
Venue used
Access, adoption, and broker incentives all matter
The model identifies barriers to access. It does not prescribe an optimal regulatory tariff.
Reserve · Literature
Where this paper fits
Networks & liquidity
Katz & Shapiro (1985, 1986); Pagano (1989)
Expectations, critical mass, concentration
Exchange competition
Parlour & Seppi (2003); Foucault & Menkveld (2008)
Fragmentation and competition for order flow
Selling market data
Cespa & Foucault (2014); Brogaard, Brugler & Roesch (2024);
Hendershott, Rysman & Schwabe (2025)
Data pricing, subscriptions, and trading
Strategic pricing
Salop & Scheffman (1983, 1987)
Cost-raising strategies
Here: separately priced data gate the liquidity externality
Reserve · Global exclusion
What “exclusion” means
Eliminates positive challenger-data adoption
The challenger may still execute non-subscriber orders
Best response to a fixed challenger offer
Within either fixed or metered pricing
High adoption selected whenever it exists
Reserve
Every exchange is a monopolist for its own data. Yet data is how venues compete.
dominant
venue
traders
challenger
everyone sees the dominant book
who buys the challenger's data?
A challenger's data is worth buying only if others buy it too: they post the orders that make its book worth seeing.
Reserve
Challenger data fee
share of traders who buy challenger data \(\alpha\)
viability bound \(\bar d_c\)
challenger fee \(d_c\)
\(\alpha_L\)
\(\alpha_H\)
0
adoption unravels
adoption grows
filled dot: stable
open dot: unstable (tipping point)
Reserve
The fee that adoption can sustain
Challenger-data adoption
Sustainable
data fee
0
More users make
the data more useful
The next buyer trades less:
less benefit from the feed
Curve: willingness to pay of the marginal subscriber — a different trader as adoption grows.
Reserve
Does the squeeze pay? Start at the data-revenue maximiser
profit falls
profit rises iff some active traders do not subscribe, and the challenger's share of subscriber flow \(\sigma\) exceeds a threshold \(\hat\sigma\)
Why: higher \(m\) lowers adoption; volume moves from subscribers (route \(\sigma\) to \(c\)) to non-subscribers (route \(\lambda<\sigma\)).
\(\hat\sigma\) is low when \(\lambda\) and \(\delta\) are small, and falls as the challenger nears its viability bound.
The toll pays when the challenger is taking sufficient trading income from the dominant venue's subscribers.
Reserve
Traders lose when adoption fails
surplus at cost \(c_0\)
adoption
strictly higher, for any fixed dominant tariff (revealed preference)
gain on the intensive margin \(= \dfrac{\delta\,\alpha_H^3}{2}\)
The toll transfers surplus from traders to the dominant venue and destroys some on the way.