Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Prudent Valuation & Model Risk Quantification

Kivilo, Erik and Olofsson, Carl (2014) FMS820 20142
Mathematical Statistics
Abstract (Swedish)
This paper is a master's thesis by Erik Kivilo and Carl Olofsson at the Faculty
of Engineering (LTH) at Lund University. The research and writing
took place during the fall of 2014 under the supervision of Magnus Wiktorsson
at the Division of Mathematical Statistics at Lund University, and Per
Thastrom at EY, Copenhagen.
The thesis concerns prudent valuation of fair-valued nancial instruments
under current regulation on capital requirements for credit institutions and
investment rms within the EU. The aim is to explain the concept of prudent
valuation, and develop statistical methods for the calculation of additional
valuation adjustments (AVAs) required by the regulations. As there has
been little focus on model risk in previous... (More)
This paper is a master's thesis by Erik Kivilo and Carl Olofsson at the Faculty
of Engineering (LTH) at Lund University. The research and writing
took place during the fall of 2014 under the supervision of Magnus Wiktorsson
at the Division of Mathematical Statistics at Lund University, and Per
Thastrom at EY, Copenhagen.
The thesis concerns prudent valuation of fair-valued nancial instruments
under current regulation on capital requirements for credit institutions and
investment rms within the EU. The aim is to explain the concept of prudent
valuation, and develop statistical methods for the calculation of additional
valuation adjustments (AVAs) required by the regulations. As there has
been little focus on model risk in previous regulations, the main objective is
to quantify model risk AVA in a way that is compliant, using current research
on prudent valuation and model risk.
The method suggested in this paper captures the instantaneous valuation
uncertainty related to model risk as dened in the regulation. The rst
step of the method is to dene a group of plausible models and calibration
approaches for the instrument type. Each combination of model and calibration
is then assigned a probability weight based on the number of parameters
in the model, and a measure of t that includes all available market data.
When prices for the instrument have been calculated for all the dierent
models and calibrations, the probability weights are used to form a cumulative
price probability distribution for the instrument. The method is to our
understanding in line with the current regulation, and should according to
us hold some advantages compared to other proposed methods. (Less)
Please use this url to cite or link to this publication:
author
Kivilo, Erik and Olofsson, Carl
supervisor
organization
course
FMS820 20142
year
type
H2 - Master's Degree (Two Years)
subject
language
English
id
4888910
date added to LUP
2014-12-22 11:50:34
date last changed
2016-02-04 04:01:19
@misc{4888910,
  abstract     = {{This paper is a master's thesis by Erik Kivilo and Carl Olofsson at the Faculty
of Engineering (LTH) at Lund University. The research and writing
took place during the fall of 2014 under the supervision of Magnus Wiktorsson
at the Division of Mathematical Statistics at Lund University, and Per
Thastrom at EY, Copenhagen.
The thesis concerns prudent valuation of fair-valued nancial instruments
under current regulation on capital requirements for credit institutions and
investment rms within the EU. The aim is to explain the concept of prudent
valuation, and develop statistical methods for the calculation of additional
valuation adjustments (AVAs) required by the regulations. As there has
been little focus on model risk in previous regulations, the main objective is
to quantify model risk AVA in a way that is compliant, using current research
on prudent valuation and model risk.
The method suggested in this paper captures the instantaneous valuation
uncertainty related to model risk as dened in the regulation. The rst
step of the method is to dene a group of plausible models and calibration
approaches for the instrument type. Each combination of model and calibration
is then assigned a probability weight based on the number of parameters
in the model, and a measure of t that includes all available market data.
When prices for the instrument have been calculated for all the dierent
models and calibrations, the probability weights are used to form a cumulative
price probability distribution for the instrument. The method is to our
understanding in line with the current regulation, and should according to
us hold some advantages compared to other proposed methods.}},
  author       = {{Kivilo, Erik and Olofsson, Carl}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Prudent Valuation & Model Risk Quantification}},
  year         = {{2014}},
}