CPD EP28 - When Models Meet Behaviour: The Psychology of Treasury Risk with Svetlana Kardan

When Sophisticated Models Create False Confidence

Banks have never had more sophisticated tools to measure financial risk.

Interest rate risk models, liquidity forecasts, stress testing frameworks and regulatory ratios provide an extraordinary amount of information about balance sheet exposures.

Yet sophisticated analysis does not necessarily produce better decisions.

A model can be technically sound, supported by extensive data and accepted throughout an organisation, while still creating a dangerous illusion of certainty.

That was one of the most striking themes in my conversation with Svetlana Kardan, former HSBC Group Head of Interest Rate Risk in the Banking Book (IRRBB), author and creator of Psychology of Treasury.

Svetlana has spent more than two decades working across financial markets, Treasury and Asset and Liability Management (ALM). Her experience has led her to explore a fundamental question: why do intelligent professionals, working within sophisticated financial institutions, still make poor decisions about their balance sheets?

The answer is not always found in the models themselves.

It is often found in the assumptions people make, the confidence they place in analytical outputs, and the organisational behaviours influencing how information is interpreted.

The implications extend well beyond Treasury.

They challenge how risk functions, executive committees and boards understand uncertainty, exercise judgement and make decisions.

The Psychology Behind Treasury Risk Management

Treasury is traditionally viewed as a quantitative discipline, supported by sophisticated models, regulatory metrics and extensive financial analysis.

Measures such as the Liquidity Coverage Ratio (LCR), Net Stable Funding Ratio (NSFR), Economic Value of Equity (EVE) and Net Interest Income (NII) sensitivities provide essential information about a bank's financial risk exposures. Yet their usefulness depends not only on the quality of the calculations, but also on the assumptions underpinning them.

Svetlana explained that, in her experience of Interest Rate Risk in the Banking Book (IRRBB), behavioural assumptions can account for 30% to 50% of modelled NII sensitivities. These assumptions reflect expectations about how customers, markets and institutions will behave under different conditions. They may include:

• Customer deposit behaviour: How quickly depositors withdraw funds or respond to changes in interest rates.

• Loan and mortgage behaviour: How customers adjust borrowing, repayment or refinancing decisions as market conditions evolve.

• Pricing assumptions: How changes in market interest rates are passed through to customers and products.

• Balance sheet dynamics: How customer and institutional behaviour influences the composition and sensitivity of assets and liabilities over time.

These assumptions are necessary to model financial risk, but they introduce uncertainty that cannot be eliminated through greater computational sophistication. Understanding a reported sensitivity without examining the behavioural expectations behind it provides an incomplete picture of the underlying exposure.

This is where Svetlana's Psychology of Treasury offers a different perspective. Her work examines why intelligent professionals, operating within well-resourced financial institutions, can still make poor balance sheet decisions despite having access to sophisticated analytical tools and established governance frameworks.

The challenge is not necessarily that models are inaccurate or that information is unavailable. A framework may be correctly implemented, a model may operate as designed, and a committee may receive technically accurate reporting, yet the resulting decision can still be flawed. Behavioural biases, organisational incentives, professional experience and the way information is presented all influence how individuals interpret analytical outputs and determine appropriate action.

For Treasury professionals, risk leaders and boards, this raises a fundamental question: are we genuinely understanding the risks reflected in our models, or are we placing too much confidence in the numbers they produce?

Ultimately, effective Treasury risk management requires more than quantitative expertise. It demands an understanding of the human behaviours that shape financial assumptions, influence collective judgement and determine how organisations respond when reality differs from expectations.

The Dangerous Illusion of Model Precision

One of the most important distinctions in risk management is the difference between precision and certainty.

A model may calculate an exposure to several decimal places, but that precision does not make its assumptions equally reliable.

Svetlana described how this distinction can disappear when model outputs reach senior decision-makers.

A calculated sensitivity becomes a number in a committee paper. The number is discussed, reported and eventually treated as though it represents an established fact.

The estimate acquires an authority that its underlying assumptions may not justify.

This creates a subtle but important shift.

Instead of asking what the model tells us about possible outcomes, decision-makers begin treating the output as what will happen.

The danger becomes greater when a model has performed well historically.

Successful models develop credibility. Their assumptions become embedded in governance processes, reporting and organisational expectations.

Over time, challenging those assumptions can feel like challenging the competence or judgement of the people who developed them.

Svetlana links this behaviour to normalcy bias, where historical experience creates expectations about future conditions.

But a model performing well in the environment in which it was calibrated tells us relatively little about how it will behave in circumstances it has never encountered.

For boards and risk committees, this raises an important governance challenge.

Are we being presented with an estimate, or have we gradually learned to treat an estimate as a fact?

The distinction matters because confidence in the model can become a substitute for understanding the uncertainty it is intended to represent.

Reverse Stress Testing: Challenging What We Think We Know

One of the most interesting parts of our conversation concerned reverse stress testing and its role in challenging assumptions that might otherwise remain unquestioned.

Traditional stress testing typically begins with a defined scenario and examines its potential consequences. Reverse stress testing approaches the problem differently, starting with an outcome severe enough to threaten the organisation and working backwards to identify the circumstances that could produce it.

Svetlana values this approach because it encourages decision-makers to move beyond familiar scenarios and question the assumptions underpinning their models, contingency plans and proposed management actions. Rather than focusing exclusively on what is considered likely, it forces organisations to examine how their expectations could prove wrong.

This is particularly relevant to liquidity risk management, where institutions may identify numerous actions they could take during a crisis without sufficiently examining whether those actions would remain viable under severe conditions.

Reverse stress testing can help expose weaknesses in:

• Underlying assumptions: Customer behaviour, market liquidity and funding availability may evolve differently from historical experience.

• Management actions: Proposed responses may be ineffective, insufficient or impossible to execute within the available timeframe.

• Dependencies and constraints: Several management actions may rely on the same resources, market conditions or counterparties, limiting their effectiveness when needed most.

• Decision-making: Delayed escalation, uncertainty over responsibilities or excessive confidence in model outputs may prevent timely intervention.

The existence of a management action is therefore not evidence that it will work. Its effectiveness depends on the circumstances in which it must be executed, the assumptions supporting it and the organisation's ability to act under pressure.

Svetlana also described how she approaches situations where her professional judgement conflicts with model outputs. Rather than simply asking whether the model is correct, she considers what would need to be true for it to be wrong. This shifts the discussion away from defending a particular result towards understanding the conditions under which its conclusions might no longer hold.

Importantly, the same scrutiny must apply to professional judgement. Experience can provide valuable insight, but it can also introduce biases and reinforce assumptions shaped by previous market conditions. Neither a model nor an experienced practitioner should be treated as an unquestionable source of authority.

For risk leaders, the broader lesson is that reverse stress testing should do more than demonstrate compliance or produce another set of scenarios. Used effectively, it can strengthen organisational understanding of uncertainty, reveal weaknesses in planned responses and create a more constructive dialogue between quantitative analysis and professional judgement.

Why Liquidity Resilience Cannot Be Demonstrated on Paper

Financial institutions invest considerable effort in preparing liquidity contingency plans, recovery arrangements and crisis management procedures.

But documentation does not necessarily demonstrate operational capability. Svetlana made this distinction particularly clear when discussing liquidity crises.

Under acute pressure, the decision-making environment changes rapidly.

Time horizons shorten. The number of people involved in decisions may decrease. Emotional pressure increases, while the opportunity to gather additional information becomes more constrained.

Individuals who normally demonstrate strong strategic thinking may find themselves focused almost entirely on the next few hours. The effectiveness of the response depends heavily on what the organisation has prepared and practised beforehand.

Svetlana captured the distinction succinctly:

"The plan that was never exercised is just a document. For me, it's not a capability."

This observation extends beyond Treasury.

A liquidity contingency plan, crisis management framework or operational resilience strategy only becomes meaningful when the organisation can demonstrate that the people, processes and management actions will operate under realistic conditions.

That requires testing assumptions, exercising decision-making and validating whether proposed actions can be executed within the available time.

It also requires recognising that a liquidity crisis does not necessarily develop at a pace convenient for governance processes.

The ability to act cannot be established solely through documentation.

It must be developed through practice.

When Waiting for Better Information Becomes the Risk

One of the recurring challenges in risk management is deciding when sufficient information is available to act. Under uncertainty, requesting additional analysis can appear prudent. Yet waiting can also become a way of avoiding an uncomfortable decision.

Svetlana offered a practical test. When someone argues that more information is needed, they should be able to identify precisely what is missing and explain how receiving it would change the decision. If neither can be articulated, the organisation may not genuinely be waiting for information. It may simply be delaying action.

This distinction becomes particularly important during fast-moving events, when the cost of waiting is rarely neutral. Market conditions evolve, options disappear and the consequences of inaction can become progressively more severe.

In crisis simulations, I have frequently observed executives reach a point where the decision becomes uncomfortable enough that they instinctively seek more information. Yet the scenario continues to evolve, regardless of whether they are ready to act. The organisation does not control the pace of the crisis.

Effective risk leadership therefore requires more than analytical capability. It requires recognising when additional information will genuinely improve a decision and when the pursuit of certainty has become an obstacle to action.

How Behavioural Biases Influence Banking Decisions

Svetlana identified three behavioural biases that become particularly dangerous when Treasury decisions are made under pressure.

Sunk cost bias can encourage decision-makers to maintain a deteriorating position because reversing it would require acknowledging that the original decision was wrong.

The psychological discomfort of accepting a loss can become more influential than the financial consequences of continuing.

Groupthink can create an appearance of consensus even when individuals privately disagree with the proposed course of action.

During periods of stress, challenging the prevailing view may be interpreted as disloyalty or a failure to support the team.

The minutes may record a unified decision while concealing significant underlying reservations.

Authority bias can cause people to defer to the most senior individual rather than the person with the most relevant information.

This becomes particularly problematic during rapidly evolving events, when those closest to the situation may have a stronger understanding of emerging developments than those holding formal authority.

These behaviours reveal an important limitation of traditional governance.

Clear responsibilities, committee structures and reporting processes are necessary, but they do not automatically produce effective challenge.

The quality of decisions also depends on whether individuals can question assumptions, communicate concerns and influence outcomes.

The Same Risk Information Can Produce Different Decisions

Perhaps the most striking insight from the conversation concerned how information is presented to decision-makers and the extent to which this can influence their judgement.

Svetlana described how an Asset and Liability Committee (ALCO) might receive a single Economic Value of Equity (EVE) sensitivity figure and treat it as a definitive measure of risk. Presenting a range of outcomes, together with the assumptions driving that range, creates a different discussion. Rather than accepting a single number, committee members must exercise judgement about uncertainty and consider the conditions under which different outcomes might materialise.

But the influence of presentation goes further. Svetlana explained that even the sequence in which information is presented can materially affect the committee's response. Leading with the worst-case scenario may make subsequent outcomes appear reassuring by comparison. Conversely, presenting the base case first may cause the same worst-case scenario to appear unusually extreme.

The underlying financial position and model outputs remain unchanged, yet the committee may reach a materially different conclusion depending on which information is presented first. This illustrates how framing and anchoring can influence collective judgement, even among experienced financial professionals.

The implications for board reporting and risk governance are significant. Risk reports are not necessarily neutral simply because their numbers are accurate. The selection, framing and sequencing of information can shape how decision-makers perceive uncertainty, assess potential consequences and determine appropriate action.

Boards and executives must therefore consider not only the accuracy and completeness of risk reporting, but also how its presentation influences their interpretation and decisions. A technically accurate risk report can still shape decisions in ways that its authors may not have intended.

Effective Challenge Requires More Than Disagreement

The discussion also explored the practical realities of challenging senior stakeholders, particularly when the stakes are high and the message is unwelcome.

Svetlana emphasised that expressing discomfort is not the same as presenting a credible challenge. An instinct that something is wrong may be valuable, but it needs to be supported by evidence, analysis and a clear explanation of the concern. Equally important is the willingness to challenge one's own assumptions before questioning those of others.

Once a concern has been established, persistence becomes essential. Svetlana described occasions when she had repeatedly raised concerns or removed items from ALCO agendas because she considered the underlying methodology or modelling insufficiently developed. She also stressed the importance of formally documenting challenges, ensuring that concerns are visible in committee records rather than disappearing once a discussion has concluded.

However, not every disagreement can be resolved immediately, particularly when decisions must be made with incomplete information. In those circumstances, the assumptions supporting the decision should remain explicit and subject to reassessment as new evidence emerges. This allows organisations to move forward without treating provisional conclusions as established facts.

The challenge for risk leaders is to strike the right balance between constructive scrutiny and timely decision-making. Excessive caution can create paralysis, while insufficient challenge can allow weak assumptions to become embedded in governance and operational practices.

Effective challenge is therefore not simply about being willing to disagree. It is about making uncertainty explicit, supporting concerns with evidence and ensuring that decisions remain open to revision when circumstances change.

Leadership Reflection: The Human Dimension of Financial Risk

Towards the end of our conversation, Svetlana reflected on the personal experiences that shaped her approach to leadership and risk management. Her fascination with human behaviour began long before her Treasury career, through childhood curiosity, a love of history and an interest in Sherlock Holmes. What attracted her was not simply solving mysteries, but understanding people's motivations, fears and decisions.

Working across different countries, cultures and banking environments subsequently reinforced the importance of patience, listening and understanding different perspectives. Svetlana also reflected on how both her professional experience and raising four children helped develop her resilience. In financial markets, where conditions can change rapidly and mistakes have immediate consequences, she learned that panic rarely improves a decision. Remaining calm is not necessarily an innate characteristic, but a capability developed through experience.

Her advice to women entering financial markets was equally grounded in personal experience. She emphasised the importance of developing technical competence, building confidence and taking opportunities without waiting for permission. Reflecting on her own career, she recognised that some of her most formative experiences came from stepping into roles or situations where the path forward was not clearly defined. She also highlighted the importance of mentoring and supporting others who may not have benefited from the same encouragement and opportunities.

These reflections bring the conversation back to its central theme. Treasury risk management may rely on sophisticated quantitative analysis, but its effectiveness ultimately depends on people. Models are designed around assumptions, their outputs require interpretation, and decisions must be made by individuals whose experiences, incentives and behaviours influence how they respond to uncertainty.

For boards, risk leaders and Treasury professionals, the challenge is therefore not to choose between technical sophistication and human judgement, but to understand how they interact. Strong analytical capabilities are essential, yet they cannot compensate for an environment where assumptions go unchallenged, uncomfortable information is dismissed or decision-makers mistake precision for certainty.

Effective risk management requires both analytical rigour and behavioural awareness. The strongest organisations are those that make assumptions visible, encourage meaningful challenge and develop the confidence to act when certainty is unavailable.

---

LISTEN TO THE FULL EPISODE

---

More About the Interview with Svetlana Kardan

Why do sophisticated financial models still lead to poor decisions?

In this CPD-accredited episode of the RiskMasters Podcast, I speak with Svetlana Kardan, former HSBC Group Head of IRRBB, author and creator of Psychology of Treasury, about Treasury risk management, behavioural biases and decision-making under uncertainty.

Drawing on more than two decades of experience across financial markets and Asset and Liability Management, Svetlana explains how assumptions influence financial risk models, why historical success can create false confidence, and how organisational dynamics affect the decisions made by Treasury professionals, executive committees and boards.

We also explore reverse stress testing, liquidity crisis preparedness, the influence of information framing on ALCO decisions, and the importance of effective challenge.

Although the conversation focuses on Treasury, its lessons extend to risk management, financial governance, organisational resilience and leadership across financial services.

🎯 What You Will Learn

By listening to this episode, you'll gain practical insights into:

• Understanding how behavioural assumptions influence Treasury risk management and IRRBB.

• Recognising the difference between model precision and genuine certainty.

• Using reverse stress testing to challenge assumptions and management actions.

• Identifying behavioural biases that undermine financial decision-making.

• Strengthening liquidity crisis preparedness through practical testing.

You will also explore how risk reporting influences ALCO decisions, why waiting for information can become a risk, and how evidence-based challenge improves governance.

🕒 Episode Highlights

01:21 – From financial markets to Treasury leadership

How curiosity about human behaviour shaped Svetlana's career.

07:14 – The psychology behind Treasury risk

Why behavioural assumptions can materially influence IRRBB sensitivities.

11:31 – The illusion of model precision

How sophisticated financial models can create misplaced confidence.

13:40 – When successful models become dangerous

Why historical performance can reinforce normalcy bias.

15:41 – Reverse stress testing and professional judgement

Challenging models by examining what could make their assumptions fail.

18:30 – Liquidity crisis preparedness

Why an untested contingency plan is not a demonstrated capability.

21:23 – When waiting for information becomes a risk

Recognising when the search for certainty delays necessary decisions.

23:02 – Behavioural biases in banking

How sunk costs, groupthink and authority influence decisions under pressure.

27:25 – How ALCO reporting shapes decisions

Why identical risk information can produce different committee responses.

29:22 – Effective challenge and governance

Building credible challenge through evidence, persistence and documentation.

32:15 – Leadership and women in financial markets

Svetlana's reflections on confidence, mentoring and career development.

💡 Key Insights

Five principles emerge from the discussion:

• Sophisticated models do not eliminate uncertainty, and precision should never be mistaken for certainty.

• Historical success can make established assumptions more difficult to challenge.

• Risk management capabilities must be tested under realistic conditions, not merely documented.

• Behavioural biases and information framing can materially influence governance decisions.

• Effective risk leadership combines technical expertise, independent judgement and persistent challenge.

🎧 Listen Now On

• Spotify: https://open.spotify.com/episode/11lxf3Ea16Wmq2lMb7PPJB?si=-xRirP4jT62ZkKkaGuLDyg

• Apple Podcasts: [Insert Episode 28 URL]

• Other podcast platforms via RiskMasters.

👤 About Svetlana Kardan

Svetlana Kardan is a senior Treasury and balance sheet risk professional, author, educator and creator of Psychology of Treasury.

With more than 20 years of experience across international banking and financial markets, she has held senior positions spanning Asset and Liability Management, interest rate risk, liquidity and funding, foreign exchange, stress testing and Treasury governance.

Previously Group Head of IRRBB at HSBC, Svetlana combines technical financial risk expertise with a strong interest in leadership, behavioural psychology and organisational decision-making.

Through her writing and educational work, she explores why technically sophisticated financial institutions can still make poor decisions and how understanding human behaviour can strengthen Treasury risk management.

Svetlana on LinkedIn: https://www.linkedin.com/in/svetlana-kardan/

📚 Related Resources

Explore more insights on risk management, organisational resilience and decision-making:

🎓 Download Your CPD Certificate

After listening to the episode, access your CPD certificate through The CPD Group: TBD