Insights
24 technical articles across banking and prudential risk, insurance and actuarial modelling, marine and geopolitical risk, climate and catastrophe risk, quantitative methods and model governance.
Artificial intelligence has not created a new category of risk governance. It has stressed the existing one, by increasing the number of decision-influencing systems faster than most institutions have…
Monte Carlo simulation estimates the distribution of an outcome by generating many random realisations of its inputs. It is the standard technique wherever a problem has too many interacting sources o…
Machine-learning models are subject to the same validation obligations as any other model, with additional attention required in four areas: data leakage, stability, explainability and bias. The princ…
An economic scenario generator produces internally consistent simulated paths for interest rates, inflation, equity returns, credit spreads and exchange rates. It underpins market-risk capital, asset …
Catastrophe models are the basis on which insurers price peril-exposed business, size reinsurance and hold capital. Their outputs are also part of the reason cover is becoming unaffordable in some loc…
Security conditions do not drift smoothly; they shift between qualitatively different states. A hidden Markov model formalises this by assuming an unobserved state variable that changes according to t…
Flood-risk modelling converts physical hazard into financial loss through a chain of hazard, exposure, vulnerability and financial modules. Each link introduces uncertainty, and flood is unusually dem…
A Hawkes process is a point process in which each event temporarily raises the probability of further events. That property makes it a natural candidate for conflict and security incidents, which are …
The liquidity coverage ratio and net stable funding ratio are standardised minimum tests. They are useful for comparability and insufficient for management, because they use prescribed outflow assumpt…
Global shipping depends on a small number of narrow passages. For a marine insurer this creates a structural accumulation problem: many independently underwritten risks that are not independent at all…
IFRS 9 replaced incurred-loss provisioning with a forward-looking expected-credit-loss model. The mechanics rest on three parameters — probability of default, loss given default and exposure at defaul…
Quantitative geopolitical risk models attempt to express political and security conditions as parameters an insurer can price and capitalise. They offer consistency and speed that narrative assessment…
IFRS 17 changed how insurance contracts are measured and how profit emerges. Because management responds to what it is shown, a change in what the accounts reveal is also a change in how the business …
The internal capital and liquidity adequacy assessment processes exist to answer two questions a board should be able to answer anyway: do we hold enough capital for the risks we actually run, and can…
The own risk and solvency assessment is the process through which an insurer forms and documents its own view of the risks it faces and the capital needed to support them over its planning horizon. It…
Credit-spread risk in the banking book is the risk that changes in the market price of credit risk and of market liquidity affect the value or earnings of non-trading positions, separately from change…
Extreme-value theory provides a principled basis for estimating the probability of losses larger than anything in the observed record. Its appeal is that the limiting distributions are derived from th…
Marine war-risk pricing translates a political and security assessment into a per-transit premium. The chain runs from exposure definition through event probability and severity to accumulation, expen…
Interest-rate risk in the banking book is the exposure of a bank's capital and earnings to movements in interest rates arising from non-trading activity. It is measured from two perspectives — economi…
Climate-risk modelling estimates how a changing climate and the policy response to it affect the financial position of banks and insurers. It differs from conventional risk modelling in three ways: th…
Asset and liability management is the discipline of steering a balance sheet so that earnings, economic value, liquidity and capital remain acceptable together — not one at the expense of the others. …
Model validation establishes, through independent evidence, whether a model is fit for its declared purpose and under what conditions that conclusion stops holding. The framework below organises that …
Every insurance decision sits on the same chain: expected loss, uncertainty around it, capital to carry that uncertainty, a price that recovers the cost of the capital, and limits that keep the whole …
Actuarial risk modelling exists to convert uncertainty into decisions. Its output is not a forecast but a structured description of what could happen, how likely each outcome is, and what resources ar…