63
6.3. Brief description of SYMBOL
The Systemic Model of Banking Originated Losses (SYMBOL) model has been developed by JRC in
cooperation with members of academia and representatives of DG FISMA. The original article
describing the working of the model appeared in the peer-reviewed Journal of Financial Services
Research.
The core of the model is the Fundamental Internal Risk Based formula from the Basel III regulatory
framework. The Basel III Fundamental Internal Risk Based formula works on the idea that credit
assets outcomes fundamentally depend on a single factor.
This allows modelling and simulations to
be carried out very easily. The formula has two additional useful characteristics in terms of modelling:
(a) it uses a very limited number of parameters expressing the riskiness of credit assets and their
correlation; (b) it gives comparable results when used on a set of sub-portfolios of assets, each with its
own parameters, and then summing up results, or when directly considering the whole portfolio using
average parameters values.
The model thus assumes that: (a) the Basel 3 regulatory model for credit risk is correct; (b) banks
report risks accurately and in line with this model;
(c) all risks in the bank can be represented as a
single portfolio of credit risks.
It is then possible to use publicly available data on total regulatory
capital, risk weighted assets and total assets to obtain parameters representing the average riskiness of
each bank’s portfolio of assets.
Once parameters are obtained for all banks, a set of loss scenarios are simulated. In each scenario, a
number representing a realization of the single risk factor is randomly generated for each bank. To
represent the fact that banks all operate in the same economy, the risk factors are correlated between
themselves.
Given the realisation of the risk factors and the parameters above, it is possible to obtain from the
model a simulated loss for each bank in each loss scenario.
These losses can then be applied to bank
capital to see which banks “default” (i.e. exhaust or severely deplete regulatory capital) in the
simulated scenario. If the policy set-up allows for or any other loss-absorbing or re-capitalization tool
(e.g. bail-in) these can also be applied at individual bank level. Losses, interventions of other tools and
counts of defaults can then be aggregated across the whole banking sector. Moreover, given that the
simulations work at individual bank level, other characteristics of banks subject to “default” can be
tracked, such as covered deposits or total assets held.
Given a sufficient number of loss scenario simulations (hundreds of thousands to millions), it is
possible to obtain statistical distributions of outcomes for the banking sector as a whole.
R. De Lisa, S. Zedda, F. Vallascas, F. Campolongo, M. Marchesi; “Modelling Deposit Insurance Scheme Losses in a Basel
2 Framework”; Journal of Financial Services Research; December 2011, Volume 40, Issue 3, pp 123-141. First Online
November 2010. Please note that at the time of submission the acronym SYMBOL was not yet employed.
In a very simplified way: given the general situation of the economy, each asset will have a certain probability of
defaulting. By considering such probabilities of default as the expected loss conditional on the economic situation and
summing across assets it is possible to obtain an expected loss of the portfolio conditional on any economic scenario.
The capital requirement is then the loss on a particularly adverse scenario. (See also footnote 7).
When this is not the case, we need to rely on self-reported or supervisory assessments of the correction that would be
needed when moving from the current system to a Basel III compatible system. It should be noted that the original
framework of the model employed Basel II (and not III) compatible data, as this was the regulatory framework of
reference at the time.
This does not mean that other risks are not considered, simply that they can be “mapped” in credit risk terms and modelled
using the same framework.
Other parameters are fixed at the default levels set in the regulation.
It should be noted that SYMBOL is a “purely static” model. Losses are all realized (or known) at the same point in time for
all systems’ participants and banks do not dynamically react to events.
It is important to stress that, though the model simulates losses at individual bank level, individual bank results are not
deemed to be usable per se.