Allocation of Risk Capital in Financial Institutions

Financial Management (Financial Management Association), Autumn, 1999 by Francesco Saita

One might argue that the problem might be solved by setting the CaR limit for each unit greater than $100,000 but in such a way that, after the effect of internal risk offsets, total potential losses could remain within the desired $200,000 limit. This can be achieved, for instance, by considering correlations, as when total potential losses must be estimated on assets exposed to different risks. Unfortunately, while it is possible to estimate correlations among different risks, correlations among different trading desks are remarkably unstable. In fact, in the example, potential losses for the two traders due to British money market rate changes are inversely correlated; but if the currency trader had bought dollars against pounds instead of selling them, the risk would have been positively correlated. This implies that it is not possible to increase CaR limits for the two desks. Therefore, a certain degree of waste in risk-taking capacity (i.e., capital) will result from the partial overlap of risk responsibilities between the two desks.

A real solution to the problem consists of transferring each risk to a specialized unit through transfer pricing. A common case is the transfer of funds interest rate risk to the treasury. In the example, this would mean forcing the currency trader to transfer his British money market exposure to the money market desk, which would then become responsible for the net position. By transforming inadvertent hedges into deliberate ones, in a full specialization context, this form of capital waste would be eliminated. Nevertheless, it is not always possible or desirable to specialize all multi-risk units in a single risk unit.

B. Single Risk vs. Multi-Risk Units: Risk Specialization and the Transfer Pricing Issue

If risk specialization can reduce the improper use of capital-at-risk, the organization could conceivably be structured around the principle of risk specialization alone. Nevertheless, the organizational structure of large, diversified banks only partially adopts risk as the key attribute to define different divisions: it is not uncommon to find a risk-based division (e.g., capital markets) together with divisions responsible for customer segments (e.g., financial institutions) or geographical areas (e.g., North America, emerging markets) at the same hierarchical level in a bank. According to classical organizational design theory, organizational units should be formed by assembling those tasks having the highest level of interdependence (see Thompson, 1967, and Lawrence and Lorsch, 1967). The interdependence shown by many banks across geographical markets, customer segments, and product lines is such that deciding which dimension should prevail when creating organizational units is far from easy (Smith and Walter, 1990). The frequency of bank reorganizations underscores these difficulties. [10]

Even without driving the entire organizational form, total risk specialization could be pursued by establishing a pervasive risk-transfer system. But even though some degree of risk specialization might be desirable, total specialization may not be. In particular, at least three obstacles to total specialization can be identified: