Regulatory and ‘economic’ solvency
standards for internationally active
banks
Comments by
Giovanni Majnoni
(The World Bank)
Basel II: An Economic Assessment
(Basel Committee on Banking Supervision, May 17-18, 2002)
Main points
• The questions addressed and the results;
• Comments on methodological
questions;
• Scope of the analysis and possible
extensions.
The questions addressed
The regulatory question:
• What level of “survival probability” is
embedded in the 8% rule?
• Or what level of bank mortality is acceptable
for bank regulators?
The economic question:
• What level of “survival probability” banks
consider when setting the levels of capital
that they effectively set aside?
• What justifies prevailing levels of capital.
The same questions could be read
Moving from Basel I to Basel II :
• is it realistic the objective of the BCBS of
keeping at the 8% level the average capital
requirement in Basel II?
• should we expect a sudden rise or fall of capital
requirements?
• …and for whom?
• how robust are the empirical relationship
between capital and insolvency levels?
The answers of the paper
•
•
•
•
For large internationally active banks the 4%
requirement of core capital is coherent with a default
ratio between 4% (three years horizon, average quality
portfolio of US banks) and 0.01% (one year, good
quality portfolio);
The amount of capital held by large G10 banks is equal
on average to 7% (equivalent to less than 0,001% default
probability over one year horizon);
The market discipline enforce an economic ratio larger
than the regulatory one;
(Implicitly) switching to a regime that envisages similar
default frequencies should not have major quantitative
effects.
Methodological contributions
•
Simulation procedure for a MTM loan portfolio
exposed to a common risk factor:
–
–
–
•
follows a simplified CreditMetrics approach.
Selects a set of relevant parameters: a representative
portfolio composition, an S&P transition matrix, an
average 50% LGD modeled as a beta distribution;
See Carey (2002) for a simulation based on a DM
approach;
Empirical measurement of the “implicit public
guarantee” element built into banks ratings in order
to assess “market” evaluation of financial strength.
…and a terminology issue
Confidence interval:
• for the paper is the “survival probability” or
“solvency standard” (1-a);
• No other reference to more traditional notion of
“confidence interval”:
–
–
No measure of the precision of the estimated parameters of
the loss distribution function (mean, percentile levels);
It is not possible to distinguish whether the different default
probability levels are significantly different among
different simulations or, alternatively, how many capital
levels are compatible with one “survival probability”.
Confidence intervals
Confidence intervals (of “confidence intervals” in the
paper terminology) grow rapidly as we measure
percentiles farther away in the tail of a distribution :
•
–
n-1/2 = std. error of the sample mean of a normal
distribution;
–
kn-1/2 = std. error of a sample percentile (k= 2.13 for the
5% percentile; k= 3.77 for the 1% percentile);
–
More so for asymmetric distributions (Kupiec, 1997).
An incidental remark: lack of quantitative
assessments of general loan loss provisions, easier
to estimate, especially where data are scarce.
Scope of the analysis and open issues
1. What if we consider loan portfolios of banks not
internationally active? Or if the authors had
considered transition matrices taken from crisis
periods (Carey, 2002)?
2. When we leave the world of “internationally active
banks” an evaluation of the adequate level of
economic capital and of market discipline becomes
much harder:
1.
2.
Rating often would not be available;
In emerging economies, when available, ratings are almost
completely determined by the “support” component
(sovereign rating).
Scope of the analysis and open issues
1. What if the transition matrix looks like the following:
Transition matrix for bank loans in Mexico
A
B
C
D
E
2.4
0.9
A
76.6 15.2 4.9
B
9.3
45.8 26.1 13.0 5.8
C
2.0
2.6
60.4 27.2 7.8
D
1.4
0.6
3.2
84.8 10.2
E
0.1
0.1
0.1
0.7
98.9
Source: Lowe and Segoviano (2002)
Conclusions
•The paper provides a useful simulation
methodology to select levels of capital
requirements;
•By concentrating on large international banks
from G10 countries provides evidence of a
market discipline tighter than regulatory
standards;
•Leaves open the question of the latitude of
market discipline and of the levels of capital
when we move away from the world of
wholesale banking.