Capítulo IV. Resultados de investigación y discusión
4.2 Condiciones de trabajo y satisfacción con actividades académicas
Given the fact that some firms have sought recognition of diversification benefits in the determination of regulatory capital requirements and the relatively poor performance of risk aggregation methods in the crisis that started in 2007, financial supervisors in many countries have been considering the extent to which they should rely on the results of firms’ risk aggregation. This section summarises supervisors’ stated views on the level of confidence that they have in, and the extent of their reliance upon, firms’ risk aggregation results, their views on prospects for future efforts in risk aggregation, and the general adequacy of risk aggregation methods during the crisis.
and insurance supervisors do not rely significantly on inter-risk aggregation methods used by firms at this time.10 However, there are exceptions, as at least one supervisor noted that a few banks calculated inter-risk diversification benefits, reflecting the banks’ abilities in this particular area of modelling. One respondent offered very conditional acceptance, if the modelling was thoroughly validated and found to be acceptable for regulatory purposes. Many countries did allow for intra-risk aggregation in assessments of risk and capital, as supervisors felt such modelling was much more advanced and useful in determining an organisation’s risk profile.
overall risk control frameworks. For banking supervisors, thi review of firms’ internal risk aggregation methods under th
Assessment Process (ICAAP) required for Pillar 2 of the Basel II framework. However, most of these modelling efforts remain “work in progress,” and best practices have yet to be established. Supervisors continue to meet with firms to discuss methodologies and assumptions, as well as to talk with firms about implementation and determine the degree of reliance on risk aggregation methods by the firm’s management.
10
As noted in Section 3, the distinction between inter- and intra-risk aggregation is to some extent arbitrary, and no single, clear, internationally agreed set of definitions of risk categories currently exists.
Supervisory confidence in firms’ risk aggregation results
The primary reason given by bank supervisors for their lack of acceptance was that modelling in risk aggregation was not yet sufficiently well developed as to be proven reliable. As one supervisor put it “[t]he area of inter-risk diversification is the least well understood” within risk aggregation techniques such as economic capital modelling. Many added that there was not yet a well accepted standard in this area of risk quantification. Others noted that institutions either needed to do more work in this area, or were not yet sophisticated
is still underway in preparation for Solvency II, and thus supervisors see it as premature to be taking into account the methods used by firms to aggregate risks. However, one European supervisory
k aggregation.
focused on evaluation of risk aggregation modelling as part of Pillar 2 or as part of other supervisory activities related to evaluation of risk management. One respondent indicated that this evaluation had been built into the supervisory process, with a combination ey would consider this area as enough to model diversification benefits using more promising approaches, such as copulas. Data was cited as a general problem. Some respondents felt there was insufficient data to calibrate and back-test such models. Others noted a tendency for historical data to underestimate periods of market stress unless it was adjusted in some manner, which is generally viewed as a difficult and subjective exercise.
Some banking supervisors noted that risk aggregation modelling was primarily developed for managing a bank under “normal” conditions, or periods of little or no financial stress. Because of this, models may not necessarily be reliable for purposes of capital adequacy assessment, which typically must consider extreme tail events. In addition, some banks have credit risk as the dominant risk, with other risks in a considerably more secondary role; as a result, potential diversification benefits (other than within credit risk) would likely be small. Insurance supervisors indicated greater acceptance of firms’ modelling diversification benefit to some degree. But this acceptance was conditional, as most felt this area was still in development and required considerably more effort to develop and validate. For insurance firms in most European countries, development of risk aggregation practices
authority has encouraged insurance firms to calculate diversification benefits, as they believe this exercise is an important part of understanding and disclosing actual risk levels. One response indicated that insurance regulators are moving from a balance sheet approach to a risk-based framework, although this is in very early stages of development.
Supervisory views with respect to future efforts in risk aggregation
In terms of future efforts to achieve greater comfort with aggregation methods, supervisors again differ. Supervisors in some countries are only in the early stages of understanding how such modelling can be applied, discerning methods that appear more promising, and determining if institutions would benefit from enhanced focus on modelling ris
At this time, these supervisors do not rely on risk aggregation approaches, and have no specific timeline for relying on such methods. One noted that neither they nor the major firms they supervise rely on such methods; both parties are only in very early stages of discussing this general area of risk quantification. Risk aggregation models could be accepted for use at some time in the future, but supervisors may ultimately reach a decision that such approaches are not sufficiently reliable or worth the significant effort required.
Other supervisors were more receptive to the use of risk aggregation approaches, often as part of ICAAP, but still believe this area requires further study. To obtain comfort, these supervisors
of off-site and on-site activities. Several supervisors noted th
part of their evaluation of validation processes. Supervisors also noted they would closely evaluate the levels of diversification claimed, applying some reasonableness test to the exercise.
Adequacy of risk aggregation methods during the crisis
The general view of supervisors was that firms’ aggregation methods were inadequate during the crisis and showed several areas of weakness. One supervisor observed that models may function well for specific purposes and for specific market conditions, but might fail to estimate the capital required to survive the losses under stress events. A notable weakness was reliance on assumptions that markets would be liquid; another area of weakness was inaccurate assumptions about risk transfer. One response noted the need to include exposures not captured by traditional measures, such as off-balance sheet exposures, exposures due to explicit or implicit support, linkages between direct and indirect sources of risk due to concerns about reputation, and the potential for risks to interact during times of stress. The recent turmoil emphasised to supervisors that correlations and risk interactions may change very rapidly under stress, more rapidly than risk management processes can be
uring the crisis, many existing platforms were challenged to aggregate risk
practices” needed to be established, which likely will require considerable effort. Accordingly, many expressed a desire to encourage further dialogue and development in this area. Several supervisors noted that better risk aggregation could foster
respondent noted that in general, stress tests did not seem to be sufficiently forward-looking or adequately firm-wide. To the extent that firms did not utilise stress tests as part of their decision making processes, adjusted to incorporate such changes. Stress tests could not anticipate the evolution of interactions and the relationships between risks and second order effects, unless very severe scenarios were considered.
Supervisors noted several concerns regarding data and its application. In general, most supervisors concluded that the crisis surfaced notable data deficiencies, whether due to lack of historical data or lack of applicable data. One supervisor emphasised the need to mitigate potential modelling failures or data deficiencies by using stress testing processes of banks. One respondent noted that risk aggregation frameworks place high demands on information systems; d
exposures with greater granularity and in various combinations. IT systems were not sufficiently dynamic or flexible, and information requests by supervisors resulted in ad hoc solutions.
Drawing from these experiences, supervisors generally did not foresee dramatic changes in their reliance on firms’ internal aggregation methods. The general view was that substantial improvements and refinements in methods would be needed before supervisors would be comfortable placing significantly greater reliance on risk aggregation results. Many supervisors suggested that “best
better overall risk management.
Stress testing as an alternative
Drawing from the experience from the crisis, many responses indicated that supervisors have enhanced or emphasised stress testing. However, supervisors are aware of the limitations of stress tests and noted a range of concerns. One
there was a concern with lack of involvement by top management.