It is widely agreed that a reasonable and consistently applied method of evaluating securities lending performance would benefit both lending clients and their agents.
But significant differences of opinion over the best way to establish objective performance measurement criteria have developed within the industry.
The approach most commonly proposed is one in which performance is evaluated against a series of industry-provided averages. The principal concern of some observers is that this addresses neither the meaningful differences between lending programmes nor the level of risk undertaken. While pundits have suggested varying reasons for the conflict of views, the most likely cause is the most obvious: doing it right is very difficult.
On one level, securities lending is a relatively straightforward investment management activity, but there are unique aspects that make the simple comparison of portfolio performance to an industry index problematic. For example, income from securities lending transactions collateralised by cash can be looked at as having two components: the investment spread generated by the collateral reinvestment process, and the natural, or demand, spread generated by the lending process. Each unit of spread generated by the lending part of the process entails different, usually less, risk than a unit of spread generated by the collateral reinvestment process. The use of simplified income information does not allow for the incomparability of income generated by the lending and reinvestment activities, and thus does not present a complete picture of an agent’s lending performance. There are dozens of other factors to include in an analysis system, but these two are perhaps most crucial, and bear further discussion.
On the loan, or liability, side, the natural spread is typically a function of the demand value of the security available for lending. The market is somewhat bi-modal, however, with the special”, or high demand/low supply, component commanding significant premiums based on scarcity, and the general collateral component commanding lower premiums influenced by overnight pricing levels and funding costs. The underlying manager’s investment strategy dictates the proportion of specials in a loan portfolio. The performance differences caused by the variability in portfolio composition are confused further by the use of a variety of lending philosophies - some agents lend only specials, others maximise income by “buying” volume to increase asset side investment return, still others use a combination of both. A high-level look at income - volume times spread - cannot capture the differences in risk and return in these examples, and a portfolio owner is understandably at a loss to identify the components of performance and risk.
The asset side of the ledger, collateral reinvestment, on the surface appears to lend itself most readily to a traditional benchmarking process. It is also difficult to interpret, however, due both to the wide range of collateral reinvestment guidelines and management of the underlying portfolio. Obvious determinants of return, including asset type, maturity, duration, liquidity and credit quality, are largely dictated by the underlying portfolio owners. But in a portfolio where the primary manager’s strategy entails high turnover, opportunities to add value through lending may exist only in the investment of cash on an overnight basis. Other portfolios employing a less aggressive management style may present opportunities for higher returns in longer duration vehicles. Therefore, the level of investment returns from one lending portfolio to the next may be influenced as much by turnover in the underlying portfolio as by the client’s choice of investment guidelines. In each case, performance can be evaluated accurately only in the context of such fundamental differences.
The brief examples above only hint at the true complexity of the problem. As the structure of a lending programme is ultimately dictated by the aggregate risk/return tolerances of its underlying clients, it makes sense that performance likewise be evaluated within a construct that allows all relevant risk elements to be quantified and correlated. The linking of disparate factors can only be accomplished through the use of sophisticated tools. In our view, the value-at-risk (VAR) approach to risk and performance measurement provides an accepted platform on which to build such an unbiased system of analysis.
Criticism of the VAR approach has centred on the idea that different assumptions, data sets and calculation methodologies can produce vastly diverse results. We not only agree with this premise, we regard it as one of the long-term strengths of the overall concept: clients should be empowered to select confidence intervals, correlation and volatility data sets, and other measurement criteria that satisfy their own risk management standards and reporting requirements. The rapid evolution of technology is providing the power to calculate complex models on demand and the flexibility to perform scenario analysis under a variety of assumptions. Further, as institutional investors embrace the VAR concept in greater numbers for other risk management purposes, standardisation makes sense.
The move to standardisation of VAR methodology is already in evidence with the impending shift by the Federal Reserve to a market-based system for the computation of regulatory capital for certain US banks - an asset/liability analysis not very different from that presented by securities lending.
Risk-adjusted performance measurement using accepted portfolio management tools can be of significant value in tailoring a programme to meet individual client requirements. “