REINSURANCE · DESK / PROTODESKRDM-03

ACTUARIAL / ANALYTICS

RDM-03 creates an analytics problem before it creates a loss estimate: whether the current data, assumptions, correlation structures and scenarios are capable of representing the dependency at all.

OBSERVEDLarge-load / disconnection events

MODELLEDWider system propagation

INFERREDCross-exposure correlation

UNPROVENInsured consequence

RDM-03 / DESK ANATOMY

GUARD & CREATE

EACH ROW SHOWS A SEPARATE DECISION PATHWAY THROUGH RDM-03: WHAT THIS DESK MAY NEED TO GUARD FROM THE DEPENDENCY, WHAT THE DEPENDENCY COULD HELP IT CREATE, AND WHICH DECISION MAY CHANGE.

#	Decision area	Guard from this dependency	Create with this dependency	Decision question
01	DATA FIELDS	Current datasets may not capture load behaviour, protection logic or relevant grid attributes.	New fields or proxies could make the dependency testable without rebuilding the whole model stack.	Which missing data fields would most affect the RDM-03 determination?
02	INDEPENDENCE ASSUMPTION	Models may treat exposures as independent where they share an external grid contingency.	A dependency test could identify where independence remains defensible and where it does not.	Which independence assumptions should be challenged first by RDM-03?
03	CORRELATION STRUCTURE	Existing correlation may be calibrated to known peril or event families rather than this mechanism.	A separate correlation hypothesis can be tested before it is admitted into pricing or capital models.	What correlation structure would RDM-03 require if the dependency survives?
04	SCENARIO FREQUENCY	Frequency cannot be inferred cleanly from a small set of observed initiating events.	Explicit uncertainty can prevent false precision while still supporting scenario analysis.	What frequency statement is supportable without overstating the evidence?
05	SEVERITY PATHWAY	Insured consequence remains unproven, so severity modelling can outrun evidence.	A staged pathway can separate physical propagation from insured loss translation.	What must be established before a RDM-03 severity estimate is credible?
06	AGGREGATION	Conventional aggregation may not group accounts linked by the same external dependency.	A new aggregation key or proxy could reveal hidden concentration.	How would exposures need to be grouped to test RDM-03 accumulation?
07	MODEL VALIDATION	A new variable can create spurious precision if it is not independently testable.	Falsifiable validation criteria can keep RDM-03 outside production models until warranted.	What validation evidence would be required before RDM-03 enters a model?
08	DECISION OUTPUT	Analytics can produce an interesting signal without changing a real decision.	A decision-linked output can test whether RDM-03 matters to underwriting, portfolio or enterprise risk.	Which decision would use the RDM-03 analytic output if it proved material?

Decision area

Guard from this dependency

Create with this dependency

Decision question

DATA FIELDS

Current datasets may not capture load behaviour, protection logic or relevant grid attributes.

New fields or proxies could make the dependency testable without rebuilding the whole model stack.

Which missing data fields would most affect the RDM-03 determination?

INDEPENDENCE ASSUMPTION

Models may treat exposures as independent where they share an external grid contingency.

A dependency test could identify where independence remains defensible and where it does not.

Which independence assumptions should be challenged first by RDM-03?

CORRELATION STRUCTURE

Existing correlation may be calibrated to known peril or event families rather than this mechanism.

A separate correlation hypothesis can be tested before it is admitted into pricing or capital models.

What correlation structure would RDM-03 require if the dependency survives?

SCENARIO FREQUENCY

Frequency cannot be inferred cleanly from a small set of observed initiating events.

Explicit uncertainty can prevent false precision while still supporting scenario analysis.

What frequency statement is supportable without overstating the evidence?

SEVERITY PATHWAY

Insured consequence remains unproven, so severity modelling can outrun evidence.

A staged pathway can separate physical propagation from insured loss translation.

What must be established before a RDM-03 severity estimate is credible?

AGGREGATION

Conventional aggregation may not group accounts linked by the same external dependency.

A new aggregation key or proxy could reveal hidden concentration.

How would exposures need to be grouped to test RDM-03 accumulation?

MODEL VALIDATION

A new variable can create spurious precision if it is not independently testable.

Falsifiable validation criteria can keep RDM-03 outside production models until warranted.

What validation evidence would be required before RDM-03 enters a model?

DECISION OUTPUT

Analytics can produce an interesting signal without changing a real decision.

A decision-linked output can test whether RDM-03 matters to underwriting, portfolio or enterprise risk.

Which decision would use the RDM-03 analytic output if it proved material?

THE REINSURANCE QUESTION

WHICH ASSUMPTIONS, SCENARIOS OR DATA FIELDS WOULD HAVE TO CHANGE IF CORRELATION IS MATERIALLY DIFFERENT FROM TODAY'S TREATMENT?

This is the governing question for this decision territory. The matrix above decomposes it into the smaller consequences, opportunities and questions that can be tested independently.

THREE BUYING OPTIONS

CONSEQUENCE REPORT.
OPPORTUNITY REPORT.
OR BOTH.

01 / CONSEQUENCE REPORT

GUARD FROM THIS DEPENDENCY.

THIS DESK RECEIVES: ASSUMPTIONS, VARIABLES, SCENARIOS, DATA FIELDS AND CORRELATION TREATMENTS THAT SHOULD BE TESTED OR CHALLENGED.

€25KFIXED

02 / OPPORTUNITY REPORT

CREATE WITH THIS DEPENDENCY.

THIS DESK RECEIVES: NEW VARIABLES, SCENARIOS, MONITORING AND DEPENDENCY-AWARE ANALYTICS CAPABILITY.

€25KFIXED

03 / BOTH

GUARD + CREATE.

Two complete positions from one dependency and shared evidence base.

€40KFIXED · SAVE €10K

OTHER REINSURANCE DESKS / PROTODESKS

Accumulation / Exposure Management Portfolio / Aggregate Management Treaty / Specialty Underwriting Technical Risk / Engineering CRO / Enterprise Risk Claims / Loss Scenario Cross-Class Dependency Owner

RDM-03 evidence boundary: initiating events observed; wider propagation modelled; correlation inferred; insured consequence unproven. GUARD, CREATE and DECISION QUESTION rows are outside-in decision views, not claims about any institution's actual portfolio, models or internal position.

This desk is also published as a function of the Reinsurance surface: Actuarial / Analytics →