Financial models are often treated as objective representations of reality—structured, data-driven, and inherently reliable. In practice, they frequently create something far more dangerous: an illusion of precision. Valuations are presented with confidence, cash flows projected years into the future, and sensitivities neatly quantified. Yet beneath this apparent rigor lies a fundamental vulnerability. In today’s environment—defined by structural change, shifting economic relationships, and recurring shocks—the greatest risk may not be incorrect inputs, but the model itself.
The Evolution of Model Risk
Model risk is typically understood as the possibility of making decisions based on flawed models. Historically, this was associated with tangible and identifiable issues such as poor data quality, implementation errors, or unrealistic assumptions.
What has changed is not the definition, but the source of the risk.
Many financial models still rely on a fundamental premise: that the future will behave, at least directionally, like the past. This assumption is embedded in growth projections, margin evolution, discount rates, and even in the structure of the models themselves. Stability—or at least gradual change—is often implicitly assumed.
Recent years have challenged this premise. Relationships between key variables are no longer stable. Correlations shift, sometimes abruptly. Business environments evolve faster than models are updated. In this context, even a technically correct model can produce misleading conclusions.
When the Environment Changes Faster Than the Model
The economic landscape of the last few years illustrates this shift clearly. The COVID-19 pandemic disrupted demand patterns and operations almost overnight. This was followed by a sharp change in inflation dynamics and a rapid increase in interest rates after years of relative stability. At the same time, geopolitical tensions and supply chain disruptions introduced additional layers of uncertainty.
These are not isolated shocks affecting individual variables. They represent systemic changes that alter how variables interact with each other.
Yet many models continue to extrapolate from historical data reflecting a fundamentally different environment. The result is not necessarily an error in calculation, but a mismatch between the model and the reality it is meant to represent.
Where Models Fail in Practice
The limitations of traditional modeling approaches become particularly visible under stress.
Discounted Cash Flow (DCF) models, for example, remain a cornerstone of valuation. Their structure is robust in theory, but highly sensitive in practice. In a low-rate environment, variations in discount rates had a relatively contained impact. In the current context, however, changes in the cost of capital can significantly alter valuations—often in a non-linear way.
More importantly, the issue is not limited to the discount rate itself. Growth assumptions, terminal value calculations, and margin projections are frequently anchored in historical performance. When underlying business conditions change, these assumptions can quickly become outdated, while the model continues to produce precise outputs.
A similar limitation appears in standard sensitivity analyses. Varying one input at a time provides a controlled view of risk, but not necessarily a realistic one. In practice, adverse conditions tend to affect multiple variables simultaneously. Inflation, for instance, can compress margins, weaken demand, and tighten financing conditions at the same time. Capturing these interactions requires a more integrated approach.
Another common source of risk lies in the implicit assumption that business models remain stable. Financial models often project existing operating structures forward, assuming continuity in pricing power, cost structures, and revenue drivers. However, recent years have demonstrated how quickly these elements can change.
Even scenario analysis, which is intended to address uncertainty, can fall short. Scenarios are often incremental rather than transformational, excluding extreme but plausible outcomes.
Real-World Evidence of Model Risk
These limitations are not purely theoretical. Some of the most well-known financial losses in recent history can be traced, at least in part, to model risk.
The collapse of Long-Term Capital Management in 1998 is a classic example. Its strategies relied on historical correlations and mean reversion assumptions. During the Russian Financial Crisis, those relationships broke down, and positions that were expected to offset each other moved in the same direction, leading to substantial losses.
More than a decade later, JPMorgan Chase reported losses exceeding $6 billion in the “London Whale” incident. A revised risk model underestimated exposure, allowing positions to grow beyond what the underlying risk justified.
In the case of Lehman Brothers, valuation models used prior to the Global Financial Crisis assumed levels of market liquidity that disappeared under stress, causing a significant disconnect between model valuations and market reality.
More recently, during the COVID-19 pandemic, many corporate financial models failed not due to calculation errors, but because their assumptions became obsolete almost overnight.
Across these cases, the common thread is not technical error, but structural misalignment between the model and the environment it sought to represent.
Illustrative Cases in Corporate Finance and Valuation
Beyond high-profile market events, model risk is equally relevant in day-to-day corporate finance and valuation exercises.
A common example can be found in DCF valuations during periods of rising interest rates. Models built in low-rate environments are often updated mechanically by increasing the discount rate, while leaving other assumptions largely unchanged. This creates internal inconsistencies: higher discount rates typically reflect tighter financial conditions, yet growth and profitability assumptions may remain overly optimistic.
The result is a valuation that appears technically sound but fails to fully reflect the new economic reality. In practice, this has contributed to transactions being priced on expectations that later proved difficult to achieve.
A similar issue arises in mid-market company valuations. Businesses with historically stable performance are often modeled using extrapolated trends and moderate sensitivities. However, in an environment of rising costs, demand uncertainty, and tighter financing, small deviations can compound rapidly.
Margin compression, slower growth, and increased cost of capital can significantly impact cash flows. If these effects are not modeled in an integrated way, valuations may overestimate resilience. This has, in some cases, led to misaligned expectations in transactions and post-deal underperformance.
What makes these situations particularly relevant is that the models themselves are not necessarily incorrect. The limitation lies in their inability to capture how multiple variables interact under stress.
The Illusion of Precision
One of the most significant risks in financial modeling is not error, but overconfidence.
Well-structured models and detailed outputs can create a strong sense of reliability. Valuations are presented with precision, forecasts extend several years into the future, and sensitivities suggest risks are understood.
However, this apparent precision can be misleading. A model may be internally consistent while remaining externally fragile. Decision-makers may focus on outputs without fully questioning the assumptions and structure behind them.
In this sense, model risk is not only technical—it is also behavioral.
Towards More Resilient Financial Models
Addressing model risk does not mean abandoning financial models, but rethinking how they are built and used.
A more robust approach involves moving beyond single-point estimates and considering ranges of outcomes. It also requires designing scenarios that reflect how multiple variables evolve together, rather than in isolation.
Assumptions should be dynamic rather than static, allowing key drivers to adapt to changing conditions. At the same time, transparency is critical. Models should not function as black boxes—assumptions must be clear, regularly reviewed, and actively challenged.
Ultimately, the value of a model lies not only in its outputs, but in the quality of the thinking it supports.
Conclusion: A Tool for Thinking, Not a Source of Certainty
Financial models remain essential tools. They provide structure and discipline in complex decision-making processes. However, they should not be seen as precise predictors of the future.
In an environment defined by structural change and uncertainty, models are most valuable when used to explore different scenarios and challenge assumptions—not when they create a false sense of certainty.
Recognizing the limitations of financial models is not a weakness, but a prerequisite for using them effectively.
How We Can Help
As model risk becomes increasingly relevant, the need for robust model design and independent review continues to grow.
We support clients with:
- Financial model review and validation
- Development of tailored financial models
- Scenario analysis and stress testing frameworks
If you would like to assess the robustness of your current models or develop a more resilient framework, feel free to get in touch.
