|
Market Perception
Architecture
Market Perception
Relevance
Market Acceptance
EQE - New Tornado/Hail
 Although representing 46% of all U.S. Cat losses 1949-1993, current T/H models
have fallen well short of accurately predicting actual loss.
To help better capture the range of potential loss outcomes, EQE's model…
- Employs 800,000 stochastic events that simulate over 20 million tornadoes
and 60 million hail streaks.
- Factors both spatial and temporal clustering, capturing systems which spawn
dozens or tornadoes and hundreds of hail streaks over a 2-3 day period,
- Calibrates to claims data from several large insurers.
- Relatively quick run time given number of stochastic events.
Includes “straight-line” wind losses in TO/HA events. But like all of the models,
cannot include them in more generic events, such as cold fronts and stable air
masses.
Holborn Alternatives to Commercial Models
Curve-fitting model (based on actual reinsured losses)
Deterministic “look-a-likes”
Peak Concentration / Accumulation Studies
“Rules of Thumb” (based on premium and surplus)
 |
This is not a pipe,
A pipe has weight, volume, texture, use,
scent, taste and a history and a future
independent of this view.
A picture only has color, height and
length, and does not have a past or
future. |
 |
This is not a hurricane.
A hurricane has varying rainfalls, pressure
levels, shearing forces, embedded tornadoes,
windblown debris, and follows an
unforeseeable track to interact with land
and values with their own independent
history and future.
A model only represents limited parameters,
such as maximum sustained winds, forward
speed, radius and central pressure and a
simplified track of movement. |
This is not a 500-year loss.
A return time loss is the
actual result of the most severe
loss in a period of time. It reflects
the full physical, legal, economic
and practical realities of the loss
event, the insured values and the
market.
An estimated loss only
represents the result of a model
applied to coded exposure data
files under simplifying standard
assumptions.
|
Previous