The Hidden Price

I — Two Ways to Intermediate

When a financial firm helps a client buy or sell something, it can play one of two roles. It can act as an agent — sourcing a price in the market on the client's behalf, executing the transaction, and charging a discrete, separately stated fee for that service. Or it can act as principal — stepping in as the direct counterparty, taking the other side of the trade onto its own book, and recovering its compensation not through a fee but through the price itself.

These are not merely operational distinctions. They produce fundamentally different informational environments for the client. The agent model requires the firm to show its hand: here is the market price we obtained; here is what we charge. The principal model requires nothing of the sort. The client sees a single number. Whether that number is fair, expensive, or extraordinary is a question the presentation does not invite.

The economics can be identical. A bank that converts €10,000 at a 3% spread and an agent broker that charges a €300 commission extract the same amount from the same transaction. The difference is that only one of them has to tell the client what they paid.

II — The Mechanics of Each Model

The agent model — and the broker

The agent firm does not take market risk. It accesses an external price — from an exchange, an interbank market, or a competing counterparty — and passes that price to the client, adding its fee as a visible line item. The client can observe three things: the reference price obtained, the fee charged, and the total cost. Comparison with other providers is, in principle, straightforward.

US securities law gives this role its most precise statutory expression. The Securities Exchange Act of 1934 defines a broker as any person engaged in effecting transactions in securities for the account of others (SEC, Exchange Act §3(a)(4)). The broker is legally an agent: it acts on behalf of the client, owes duties of best execution and fair dealing, and earns a commission that must be disclosed. The broker does not take the security onto its own book; it simply arranges the trade and steps aside.

The agent model's weakness is perceptual. The presence of an explicit fee makes the transaction look more expensive than it may be. A client shown a mid-market rate of 1.2700 and a £12 commission will often perceive this as costlier than a bank offering a single rate of 1.2380 — even though the latter implies a cost of roughly £260 on the same transaction.

The principal model — and the dealer

The principal firm takes the asset onto its own balance sheet, at least momentarily, and prices the transaction to recover its costs — execution, hedging, credit, and profit margin — within the spread between its buy and sell prices. Nothing is itemised. The client receives a quote; that quote is the entirety of the disclosure.

The Exchange Act's companion definition is equally precise: a dealer is any person engaged in the business of buying and selling securities for their own account (SEC, Exchange Act §3(a)(5)). The dealer acts as principal — it is the counterparty. It does not arrange a trade; it executes one against its own inventory. Compensation is recovered in the spread between the bid and the offer, not in a separately stated fee.

The spread is not fixed. It varies by transaction size, time of day, market conditions, client relationship tier, and the firm's own inventory position. This variability compounds the opacity: even a client who suspects they are paying a spread has limited means to quantify it without access to a real-time reference price — which the principal model does not provide.

In practice, most regulated securities firms in the US are registered as broker-dealers — authorised to act in either capacity and required to disclose to the client which role they are playing on each transaction. The capacity disclosure is a critical but frequently overlooked protection: a client who does not know whether their firm acted as broker or dealer cannot assess whether the price they received was fair or the commission appropriately sized. In practice, many retail clients never read the capacity disclosure, and many firms do not surface it prominently.

Model comparison

• Role — Agent / Broker: facilitator. Principal / Dealer: counterparty.
• Compensation — Agent: explicit, separate fee. Principal: embedded in spread.
• Client sees — Agent: market price + fee. Principal: all-in price only.
• Cost visibility — Agent: transparent. Principal: opaque.
• Comparability — Agent: straightforward. Principal: requires external benchmark.
• Psychological framing — Agent: appears expensive. Principal: appears free.
• Balance sheet — Agent: no exposure. Principal: firm holds inventory.
• US statutory term — Broker (Exchange Act §3(a)(4)). Dealer (Exchange Act §3(a)(5)).

III — How European Law Draws the Lines

The US framework's clarity — broker equals agent, dealer equals principal — is notable precisely because it is not replicated in European securities regulation. MiFID II does not use the terms broker and dealer as legal categories. Instead it organises trading activity by trading capacity, a threefold distinction that maps only approximately onto the US binary.

Agency

An investment firm acts in agency capacity when it executes a client order by sourcing a counterparty in the market and charging a separate commission. No principal risk is taken; the firm's book is not involved. This corresponds closely to the US broker definition and to the agent model described in this article. For MiFID II transaction reporting purposes, agency capacity is flagged as AOTC (Agent, On Trading Capacity).

Dealing on own account

An investment firm deals on own account when it trades against its proprietary capital — taking the security onto its balance sheet as a risk-bearing counterparty. This is the full principal model: the firm is the market-maker, holds inventory, and prices the transaction to include a spread. Systematic Internalisers — firms that deal on own account frequently enough to meet ESMA's quantitative thresholds — are required under MiFID II to publish firm quotes and are subject to pre- and post-trade transparency obligations. For reporting purposes this capacity is flagged as DEAL.

Matched principal — the third category

European regulation introduces a category that has no precise equivalent in US law: matched principal trading, defined in MiFID II Article 4(1)(38) as a transaction where the intermediary interposes itself between buyer and seller, concludes two back-to-back trades simultaneously, and carries no market risk. It is sometimes called riskless principal or straight-through processing.

Matched principal is economically close to agency — the firm takes no directional risk — but legally it is a form of principal trading: the firm is technically the counterparty on both legs. It earns its compensation through a small spread between the two legs rather than through a disclosed commission. The transparency implications are therefore closer to the principal model: the client sees only an all-in price, not a reference market price plus a named fee.

This distinction matters in practice. A broker crossing network or inter-dealer broker operating on matched principal terms will report capacity as DEAL under MiFID II, not AOTC — even though the firm takes no risk. Regulators and commentators have noted that this creates reporting ambiguity: matched principal activity looks like dealing on own account in transaction data, even though the economic profile is more agent-like (ICMA, 2017).

Trading capacity taxonomy

• US: Broker — Agent for client. Commission disclosed. Exchange Act §3(a)(4).
• US: Dealer — Counterparty; trades own account. Spread embedded. Exchange Act §3(a)(5).
• EU: Agency (AOTC) — Closest equivalent to US broker. Separate commission, no principal risk.
• EU: Dealing on own account (DEAL) — Full principal; risk-bearing market-maker or SI.
• EU: Matched principal (DEAL) — Riskless principal; two simultaneous back-to-back legs. No US statutory parallel. Reported as DEAL despite no market risk taken.

A further European wrinkle: MiFID II explicitly states that the distinction between acting as agent and acting as principal does not, of itself, determine the level of investor protection owed (Busch, in Oxford EU Financial Regulation, 2017). The conduct obligations — best execution, conflicts disclosure, fair treatment — apply regardless of capacity. This is a deliberate departure from a framework in which the agent/principal split carries heavy legal freight. In European law, the capacity label is primarily a transparency and reporting category; the investor protection regime is capacity-neutral.

In the US, by contrast, the broker/dealer distinction has historically carried significant legal consequences — including different fiduciary standards, different suitability obligations, and, since the introduction of Regulation Best Interest in 2020, different conduct standards depending on whether the firm is acting as broker-dealer or investment adviser. The capacity in which a US firm acts is not merely an operational fact; it is a threshold that determines which regulatory regime governs the relationship.

IV — Why the Simple Price Feels Cheaper

The principal model's commercial longevity is not solely a function of information asymmetry. It is reinforced by the way human cognition processes costs — and the way the all-in price is specifically structured to exploit that processing.

The zero-price illusion. When no fee appears on screen or paper, the brain registers the transaction as carrying zero cost. A consumer who exchanges currency at an airport bank at a rate 3% below the interbank mid has paid the equivalent of a substantial commission — but because no commission line exists, the transaction feels free. This is not irrationality in any deep sense; it is the predictable consequence of how attention and salience operate. What is not labelled is not noticed.

Complexity aversion. The agent model presents more numbers: a reference price, a spread, a commission, perhaps a platform fee. Even when the sum of those numbers is dramatically lower than an all-in principal-model price, their multiplicity creates a perception of complexity — and, paradoxically, of higher cost. In studies of financial decision-making, consumers shown equivalent costs in disaggregated form consistently rate the transaction as more expensive than when costs are bundled into a single price (Thaler, 1985).

Absence of an anchor. In agent-model transactions, the external market price serves as an anchor against which the fee is evaluated. In principal-model transactions, there is no anchor. The quoted price is not the only possible price — it is simply the only price the client has seen. Without a reference point, the concept of "expensive" has no operational meaning.

These effects do not vanish with sophistication. They attenuate under time pressure, cognitive load, and familiarity — conditions that describe most retail financial transactions. The principal model is not merely a pricing structure. It is a behavioural architecture.

V — Foreign Exchange: The Canonical Case

Retail foreign exchange at traditional banks is the most financially significant and widespread deployment of the principal model in consumer finance. When a customer converts currency through a bank branch, an online banking portal, or a travel money desk, the bank presents a single exchange rate. There is no disclosure of the interbank mid-market rate, no quantification of the spread being applied, and no commission charged.

The compensation is substantial. On major currency pairs — EUR/USD, GBP/USD, USD/JPY — retail bank spreads typically run between 1.5% and 4% of the notional converted. On minor or exotic pairs, spreads of 5–8% are not unusual. A customer exchanging £10,000 into US dollars at a bank applying a 3% spread pays approximately £300 in implicit cost. At a specialist agent-model provider — a dedicated FX platform or a modern payments firm — the same transaction might cost £10–20 in explicit fees, with a spread of a few basis points on top.

The customer who paid £300 in invisible spread will often describe their bank as offering a "competitive rate" — because no fee was charged.

The bank customer may not only be unaware of this differential; they may actively believe they received a good deal. The absence of a fee line is cognitively processed as evidence of low cost. In the meantime, the spread — invisible, unquantified, and never named — transfers a sum that would be considered outrageous if presented as a commission.

The competitive response has come from agent-model entrants: specialist FX platforms, digital banks, and payment apps that publish their spread explicitly and charge a small flat or percentage fee. Their cost structures are dramatically lower. Their adoption has grown substantially among price-conscious consumers. And yet the traditional bank model persists, in part because its customers do not realise what they are paying.

VI — Cryptocurrency: Zero Commission, Undisclosed Spread

The cryptocurrency market has reproduced the principal model with near-perfect fidelity — and with a particularly effective marketing overlay. Major retail-facing exchanges present an execution price that differs materially from the underlying spot rate on institutional order books. The difference accrues to the platform as a spread. Round-trip costs of 0.5–2.5% are common on major assets; less liquid tokens carry spreads that can reach multiples of that.

The presentation is especially effective in crypto markets for three structural reasons. First, there is no universally accepted official reference price — no interbank mid equivalent — making spread measurement harder to benchmark. Second, the intrinsic volatility of digital assets means that pricing disparities are easily obscured in noise; a 1% spread is invisible against a 10% daily range. Third, the headline "zero commission" is technically accurate and aggressively marketed, weaponising the zero-price illusion with deliberate precision.

Some platforms layer a disclosed transaction fee on top of an undisclosed spread — a hybrid opacity that is arguably more misleading than either pure model. The fee creates a sense of transparency; the spread goes unacknowledged. The client who believes they understand their cost has in fact seen only part of it.

VII — Structured Products: The Issuer as Sole Market-Maker

Contracts for Difference and exchange-traded certificates — instruments popular in European retail markets — represent the most structurally sophisticated version of the principal model. In both cases, a bank or issuer serves as the sole market-maker: the product is not traded on a centralised exchange but priced and executed entirely by the issuing party.

For CFDs, the provider typically charges no explicit commission on the trade. Instead it applies a bid–offer spread that can range from a few basis points on liquid equity underlyings to several percent on exotic or illiquid products. Overnight financing charges — swap costs or rollovers — are levied separately but are rarely understood by retail clients as a component of total cost, let alone quantified against the return of the underlying.

Certificates — turbo warrants, knock-out products, structured notes — are priced by the issuer with a structuring margin, an embedded management fee, and an ongoing secondary-market spread, all contained within a single indicative bid and ask. The client cannot observe any of these components individually. Regulatory disclosures under the PRIIPs framework (EU, 2017) — specifically the Key Information Document — partially address this by requiring a summary cost indicator. In practice, the KID's cost disclosure methodology has been widely criticised for using estimated rather than actual historical cost data, and for a presentation format that most retail clients cannot usefully interpret.

The underlying structure is nonetheless clear: the issuer is simultaneously the product manufacturer, the pricing agent, and the execution counterparty. Each of those roles carries a margin. None is separately disclosed to the client.

VIII — The Comparison Problem

The deepest difficulty with the principal model is not that it is expensive — it is that it resists measurement. Without a common reference point, comparing a principal-model provider with an agent-model competitor is structurally almost impossible for a retail client.

The agent model provides that reference point as a matter of course: the client receives both the external price obtained on their behalf and the fee charged on top. The arithmetic of total cost is immediate. The principal model provides only the final price. Recovering the implicit cost requires access to a contemporaneous market reference — the interbank mid, the exchange last price, the institutional bid — that the client does not have and the provider has no incentive to supply.

This is not a gap that transparency initiatives can fully close, because the information asymmetry is not located in disclosure language. It is located in the structure of the transaction itself. The principal model's price is technically a disclosure — the client is told the rate. What is not disclosed is the counterfactual: what the rate could have been.

Spreads also vary in ways that make comparison across time or providers difficult even for informed clients. A bank's EUR/USD rate at 9am will differ from its rate at 4pm; its rate for a €5,000 transaction will differ from its rate for €50,000; its rate for a priority banking client will differ from its rate for a standard account holder. None of these variations are published in advance. Agent-model fees, by contrast, are typically disclosed on a tariff schedule, consistent across transactions, and visible before execution.

IX — Regulation and Its Limits

Regulators across major jurisdictions have attempted to introduce transparency into principal-model markets, with mixed results.

MiFID II (EU, 2018) requires investment firms to provide clients with ex-ante cost disclosures covering both explicit charges and implicit transaction costs, including estimated spread costs. For the first time, EU-regulated firms were required to quantify and communicate spread-based costs. Implementation has been uneven; methodology disputes have limited the quality of estimates; and the disclosure format was not designed with retail comprehension in mind. Critically, MiFID II does not mandate that a firm disclose its trading capacity to the retail client in plain terms — the capacity flag appears in transaction reporting data, not in client-facing documentation.

Regulation Best Interest (SEC, 2020) requires US broker-dealers to act in the best interest of retail customers when making a recommendation, including disclosing material conflicts of interest and the capacity in which the firm is acting. Where a firm acts as principal — as dealer — in a recommended transaction, it must disclose that fact. The standard is higher than the prior suitability rule but falls short of a full fiduciary duty; critics note that "best interest" still permits principal transactions with embedded markups provided the conflict is disclosed, not eliminated.

The FX Global Code (BIS, 2017; revised 2021) establishes principles for transparency and fairness in wholesale FX markets, including guidance on spread disclosure and markup policies. The Code is voluntary and applies primarily to institutional participants. Retail FX markets remain substantially less regulated by its principles.

The UK Consumer Duty (FCA, 2023) requires firms to demonstrate that their products and services deliver fair value to retail clients — explicitly including the aggregate of all charges, implicit costs included. This is arguably the most demanding standard yet applied to principal-model pricing, requiring firms to demonstrate that their all-in rate constitutes fair value, not merely that it was stated. Early enforcement action under the Duty has focused on product design and complaints handling rather than spread quantification, but the direction of travel is clear.

The structural limit of all these approaches is the same: spread-based compensation is technically a price disclosure. The client was told the rate. Demonstrating that the rate was unfair requires benchmarking it against a reference market price — a comparison that regulators have been reluctant to mandate formally in retail contexts, and that providers have no commercial incentive to facilitate.

— The Cost of Simplicity

The all-in price is not simple. It is simplified — which is a different thing. Simplification here means the removal of information: the mid-market reference, the spread applied, the margin extracted. What remains is a single number that is easier to read and harder to evaluate.

The agent model's transparency looks like complexity. It is not. It is the natural form of a cost disclosure: here is what the market offers; here is what we charge; here is what you pay in total. That this presentation feels more complicated than a single number is a consequence of how successfully the principal model has conditioned expectations — not a reflection of any inherent difficulty in the information itself.

The practical implication for any client transacting in a principal-model context is the same regardless of asset class: find the reference price before you accept the offered price. For FX, the interbank mid is freely available. For crypto, institutional order books publish in real time. For structured products, the underlying exchange price of the referenced asset provides a floor. The difference between the reference and the offered price is the cost you are paying — whether or not it is labelled as such.

Until clients routinely demand that reference, and until regulation routinely requires it to be provided, the spread will continue to function as the financial industry's most durable form of hidden fee — embedded in the one number clients are shown, and invisible in the one number they are not.