Abstract
A compliance risk does not come into being because an employee mishandles a process. It comes into being when management chooses markets, products and clients: that choice sets which norms reach the firm. An operational act or omission is only the occurrence that meets a norm’s condition and stands as its breach. The confusion is inherited from Basel, where it was defensible: the operational-risk definition is a capital-and-loss-booking instrument, it includes legal risk — the sanction at the end of the chain — and excludes strategic risk, the place the exposure is created. Because the last link was booked as operational, the link before it was pulled in after it. The result is a chain filed under its channel, with its origin outside the definition.
A Claim Worth Examining
The sentence is everywhere: compliance risk is a part of operational risk, or arises “on the ground of” operational risk. It sounds harmless. It decides who owns the risk, where the appetite for it is set, and which committee sees it first.
It is also backwards. It treats the channel through which a compliance breach becomes visible as if it were the reason the breach was possible. To see why, start where the rules start.
The Business Model Decides Which Rules Apply
The Basel Committee defines compliance risk as “the risk of legal or regulatory sanctions, material financial loss, or loss to reputation a bank may suffer as a result of its failure to comply with laws, regulations, rules, related self-regulatory organisation standards, and codes of conduct applicable to its banking activities” [1, ¶3]. The last four words carry the argument. The rules that apply are a function of the activities, and the activities are chosen.
- Markets and jurisdictions. A firm that decides to acquire US clients across the border places itself under US securities and tax law. A joinery that serves customers in one Swiss canton does not have these risks in its universe at all — not managed well, not managed badly, simply absent.
- Products and services. Structuring derivatives or holding crypto assets in custody brings investor-protection, anti-money-laundering and prudential obligations with it. A software house selling subscriptions inherits none of them. The Committee says as much: compliance rules “may extend to tax laws that are relevant to the structuring of banking products or customer advice” [1, ¶4].
- Clients and distribution. Private banking for politically exposed persons produces an entirely different compliance landscape from domestic retail banking — even inside the same group, on the same core banking system, run by the same operations team.
The last example is the telling one. Same processes, same people, same systems — and a completely different compliance exposure. If compliance risk grew out of operational risk, identical operations would carry identical compliance risk. They do not, because the variable that differs is not operational.
In the vocabulary of the compliance-ontology essay [4], the business model fixes the Anwendungsbereich — the scope under which rights, duties and prohibitions reach the firm. That scope belongs to the meta-level above the norms, and it holds before any event occurs. Layered Event Modelling states the same thing formally: the scope of a norm “holds before any event occurs and licenses which normative edges can occur at all” [6, §6.1]. The business model is where that licence is issued.
Where the Pain Starts: Basel
The habit of calling compliance “operational” did not come from nowhere. It comes from the top of the stack, the capital framework that most bank risk taxonomies still descend from. Basel II defined the term in a single paragraph [2, ¶644]:
“Operational risk is defined as the risk of loss resulting from inadequate or failed internal processes, people and systems or from external events. This definition includes legal risk, but excludes strategic and reputational risk.”
Footnote 97 says what legal risk means: it “includes, but is not limited to, exposure to fines, penalties, or punitive damages resulting from supervisory actions, as well as private settlements.” The current consolidated Basel Framework carries the sentence and the footnote unchanged [3, OPE10.1].
Read the definition as a causal map and three things stand out.
It includes the end of the chain. Fines, penalties, settlements: these are not the breach of a norm, they are what a supervisor, a prosecutor or a counterparty does after the breach. Legal risk is the last link, and the one where the money is finally booked.
It excludes the beginning. Strategic risk is out by definition. Entering a market, launching a product, accepting a client segment — the decisions that set which norms reach the firm — sit in exactly the category the definition leaves out.
Its event types name the business model. The loss-event taxonomy that accompanies the definition [2, Annex 9] has a level-1 category called Clients, Products & Business Practices: clients, products and business practices are the three levers of a business model. Its definition covers losses “from an unintentional or negligent failure to meet a professional obligation to specific clients … or from the nature or design of a product.” The nature or design of a product is a business decision. The scheme files it as an operational loss.
None of this is a mistake for what Basel was built to do. The framework collects loss data and sets capital against it. For that job it sorts losses by how they materialise and are booked — a settlement perspective, not a causal one, and it never claimed to be anything else. The analysis of the seven event types on tlctc.net tests each of them for where it sits between cause and consequence [8]. The problem starts when the booking scheme is read as a map of where risk comes from.
The Inference Error
Basel never said compliance risk is operational risk. Its compliance paper is careful: some banks “may wish to organise their compliance function within their operational risk function, as there is a close relationship between compliance risk and certain aspects of operational risk. Others may prefer to have separate compliance and operational risk functions” [1, ¶8]. That is an organisational option, not a taxonomy.
The taxonomy came by inference. A legal risk event presupposes a compliance risk event: no sanction without a breach. Legal risk, the consequence, was filed under operational risk. So the breach it presupposes was pulled in after it. The pull felt natural because a compliance breach nearly always becomes real through something operational: a person acts, a system does or fails to do something, a control has a gap. The loss lands in an operational event type, and after enough years of loss-data collection the event type starts to look like the cause.
Written as a chain, the error is easy to see:
The three links are of different kinds. The scope is not an edge at all: it is the standing condition under which the norm applies, set by a strategic decision. The step from the act to the violation is normative: the act meets the norm’s condition and stands as its breach, at once, with no interval and nothing that could be interrupted. Only the last step is causal: someone acts on the breach, and that takes time — sometimes years [6, §6].
Basel’s definition takes the causal tail (LR) as operational, keeps the operational vector, and leaves the origin (strategy) outside. Filing the compliance breach under operational risk then collapses a normative link and a strategic origin into the one box that holds the channel.
Three Layers, Not One
Pulling cause and manifestation apart gives three layers, each with its own question and its own owner.
| Layer | What happens here | Example | Owner |
|---|---|---|---|
| 1 · Business model (genesis) | Markets, products and clients are chosen. This sets which norms reach the firm. | Offshore wealth management for clients in high-risk jurisdictions | Board |
| 2 · Compliance risk (norm) | The risk that conduct stands as a breach of a binding norm. | Cross-border tax enforcement; anti-money-laundering law | Compliance |
| 3 · Operations (vector and control) | Processes, people, systems and controls that are supposed to keep conduct inside the norm — or fail to. | An onboarding form left incomplete; a monitoring rule that never fires | Operations, control function |
Layer 3 is real and it matters: it is where controls live and where most breaches are first observed. But it is the vector through which an exposure created in layer 1 becomes manifest. Fixing layer 3 changes how likely the exposure is to turn into a breach. Only layer 1 changes whether the exposure exists. Legal risk sits after all three: legal counsel inherits a case and cannot undo the breach.
It is the same cause-versus-outcome error that TLCTC removes from cyber risk, one floor up. “Ransomware” and “data breach” were used as threat categories for years although they name outcomes; Axiom III separates the two [7]. “Compliance is operational” names the channel and calls it the cause.
What this does not claim
It does not claim that operational controls are unimportant: onboarding checks, monitoring rules and the internal control system are where a compliance function spends most of its day. It does not claim that the Basel event types are wrong for loss collection and capital. And it does not claim that every compliance risk is strategic to the same degree: for a firm whose business touches few regulated activities, compliance risk genuinely is mostly a matter of executing well. The claim is narrower. Where the business model is built on regulatorily sensitive niches, compliance risk is a first-order strategic risk, and a taxonomy that files it under operations hides its origin.
What Follows
- Appetite belongs where the model is set. Accepting a norm regime is a board decision, taken when a market, product or client segment is approved. Record it there, with its compliance exposure stated — not discovered later in the loss database.
- The register needs a layer above the controls. A GRC tool that starts at control testing can only report on layer 3. It needs an upstream view: which norm regimes each business line has opted into, and why.
- Remediation has two levers. Better controls lower the probability that exposure turns into a breach. Only a change in the business model lowers the exposure. A plan that only pulls the first lever is treating the vector.
Cause First, Channel Second
The business model sets the scope. Operations supply the act. The norm makes it a breach. The sanction comes last.
Basel booked the last link as operational risk, and for a capital model that was defensible. Filing the whole chain there is not.
A Note on the Debacles
History is a further test of the claim, kept separate here because the cases are illustrations, not the argument. The largest compliance disasters of the past two decades were rarely caused by a badly completed form.
US cross-border private banking. UBS entered a deferred prosecution agreement with the US Department of Justice in February 2009 and paid USD 780 million [9]. Wegelin & Co., Switzerland’s oldest bank, was indicted in 2012, pleaded guilty in January 2013 and announced that it would close [10]. The line of business was serving US clients whose tax obligations lay elsewhere.
Danske Bank’s Estonian branch. Roughly EUR 200 billion in payments, a large share of them suspicious, flowed through a non-resident portfolio between 2007 and 2015 [11]; the bank pleaded guilty in the United States in December 2022, with penalties of about USD 2 billion [12]. Monitoring failed, certainly — but the portfolio existed because it was profitable, and it was profitable because of who its clients were.
In each case there were operational failures, and they are where the loss events were booked. Each was downstream of a decision to be in that business. Recording these cases as control deficiencies is accurate as far as it goes. As an explanation it misses the origin.
References
- Basel Committee on Banking Supervision (2005). Compliance and the compliance function in banks. BIS, April 2005, ¶¶3, 4, 8.
- Basel Committee on Banking Supervision (2006). International Convergence of Capital Measurement and Capital Standards: A Revised Framework — Comprehensive Version (BCBS 128). BIS, ¶644 with footnote 97; Annex 9, Detailed Loss Event Type Classification.
- Basel Committee on Banking Supervision. The Basel Framework, OPE10 Definitions and application, ¶10.1 with footnote 1 (version in force from 1 January 2023). BIS.
- Kreinz, B. (2026). Rechte, Pflichten, Verbote: Compliance als Ontologieproblem. Barnes Projects. barnes.ch
- Kreinz, B. (2026). Why NIST Is Sleeping Through a Conceptual Revolution. Barnes Projects. barnes.ch
- Kreinz, B. (2026). Layered Event Modelling: Semantic Separation as a Precondition for Explicit Causal Analysis, v1.0. Zenodo. doi:10.5281/zenodo.23438101
- Kreinz, B. (2026). Top Level Cyber Threat Clusters (TLCTC), Version 2.7 — Core. Zenodo. doi:10.5281/zenodo.23437984
- Kreinz, B. (2026). The Crux of Banks Regarding Operational Risk Management. tlctc.net, February 2026. tlctc.net
- U.S. Department of Justice (2009). UBS deferred prosecution agreement, 18 February 2009.
- United States v. Wegelin & Co., U.S. District Court, S.D.N.Y.; indictment February 2012, guilty plea January 2013.
- Bruun & Hjejle (2018). Report on the Non-Resident Portfolio at Danske Bank’s Estonian Branch. 19 September 2018.
- U.S. Department of Justice (2022). Press release on Danske Bank’s guilty plea, 13 December 2022.