Book value optimisation, risk and redistribution
File(s)
Author(s)
Herskovits, Jean
Type
Thesis
Abstract
Across three different settings, we show that if classical metrics are key to understanding
behaviours of financial actors, it is necessary to also examine others, generally
considered as secondary, to comprehend actions taken by decision-makers. Firstly,
while managers of public companies maximize investors’ utility first, they also use redistribution
to manage the company’s earnings-per-share ratio, stock price, and their
own managerial stock options’ valuation. We prove that in a frictionless setting, share
repurchases and dividends are equivalent in analogy to the classical Modigliani-Miller
theorem. We further exhibit extensions in which the utility equivalence of dividends
and repurchases is broken. Our analytical differentiation of redistribution methods
provides grounding towards explaining the various empirical redistribution behaviours
observed in markets. Secondly, we prove and display empirically that typical risk metrics
such as Value-at-Risk and Expected Shortfall mechanically grow as the number
market observations is reduced. Modern international regulation assumes the opposite
and in turn implies the double counting of risk over illiquid assets. We argue that
this accounting methodology specification has considerable ramifications, altering the
global structure through which banks calibrate their risk limits: regulated banks are
now opting for the “Standardised Approach” instead of the “Internal Model Approach”.
Finally we present a novel sovereign credit grades prediction model, which beats Nomura’s
previous internal implementations as well as, to our knowledge, all previously
published ones. It is the first to attain a sufficient level of accuracy satisfactory to financial
institutions. In particular, better modelling of credit grades entails a more precise
calibration of credit risk. Not linked to material adjustments in P&L or positions,
improved credit grades’ modelling can reduce portfolios’ uncertainty: changes to the
predictions of purely indicative sovereign credit grades have meaningful repercussions
over the haircuts and spreads observed in sovereign bond markets.
behaviours of financial actors, it is necessary to also examine others, generally
considered as secondary, to comprehend actions taken by decision-makers. Firstly,
while managers of public companies maximize investors’ utility first, they also use redistribution
to manage the company’s earnings-per-share ratio, stock price, and their
own managerial stock options’ valuation. We prove that in a frictionless setting, share
repurchases and dividends are equivalent in analogy to the classical Modigliani-Miller
theorem. We further exhibit extensions in which the utility equivalence of dividends
and repurchases is broken. Our analytical differentiation of redistribution methods
provides grounding towards explaining the various empirical redistribution behaviours
observed in markets. Secondly, we prove and display empirically that typical risk metrics
such as Value-at-Risk and Expected Shortfall mechanically grow as the number
market observations is reduced. Modern international regulation assumes the opposite
and in turn implies the double counting of risk over illiquid assets. We argue that
this accounting methodology specification has considerable ramifications, altering the
global structure through which banks calibrate their risk limits: regulated banks are
now opting for the “Standardised Approach” instead of the “Internal Model Approach”.
Finally we present a novel sovereign credit grades prediction model, which beats Nomura’s
previous internal implementations as well as, to our knowledge, all previously
published ones. It is the first to attain a sufficient level of accuracy satisfactory to financial
institutions. In particular, better modelling of credit grades entails a more precise
calibration of credit risk. Not linked to material adjustments in P&L or positions,
improved credit grades’ modelling can reduce portfolios’ uncertainty: changes to the
predictions of purely indicative sovereign credit grades have meaningful repercussions
over the haircuts and spreads observed in sovereign bond markets.
Version
Open Access
Date Issued
2024-08
Date Awarded
2024-10
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Muhle-Karbe, Johannes
Tse, Alex
Sponsor
Nomura Group (Firm)
Publisher Department
Mathematics
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)