Concepts of portfolio management; Part 2: understanding and using the model

RMA Journal, The, May, 2002 by Peter Larr, Arthur Stampleman

Part 1 of this article, published in April, presented a non-technical profile of portfolio management concepts with emphasis on their use as an important management tool. Part 2 focuses on standard models and their application to portfolio management and banking.

Using the definition of portfolio management as the creation and maintenance of the best possible asset mix, we see the model as the recipe for achieving such a mix. One widely used recipe for portfolio management can be found in the models flowing out of modern portfolio theory (MTP), which are used extensively in connection with securities.

Modern Portfolio Theory Model

Among the ingredients that MPT examines for each asset, the following three are key:

1. Rate of return, measured in terms of dividend or interest yield plus percentage market gains or losses.

2. Risk, measured by volatility of return.

3. Covariance of returns.

In the securities industry, historical data on these measures is readily available. For example, there is stock market data in extensive detail going back to 1926. In addition, the securities professional has learned how to apply sophisticated mathematics (optimization techniques and mean variance analysis, for example) to this data to find the best asset mix.

Even so, use of models has not been a simple or straightforward matter in the securities industry. If it were, all investors would act the same. Obviously, they do not.

The complexity of the mathematical techniques can lead to a range of results: Each investor has different objectives, and each has his or her own proprietary concerns. The historical numbers are only the starting point, as each portfolio manager or investor must adjust the numbers to reflect his or her outlook for the future. And no two outlooks necessarily are the same. Testimony to both the complexity and the significance of MPT is the fact that Nobel prizes have been awarded for contributions to this field of knowledge.

Securities Models in Banking?

Modeling based on MPT has been a starting point for many bankers in developing portfolio management tools. Unfortunately, working with bank loans is quite different from working with marketable securities--there are many differences and difficulties:

* Historical portfolio information and data have not been as plentiful as that available for securities.

* Loan marketability and liquidity are far more limited, slowing banker's response time.

* Valuation of bank loans is far more difficult.

* Loans generally have limited upside potential, but they can drop in value to zero because of write-offs.

* Mean variance analysis techniques are not readily applicable to bank loans.

* Benchmarking commercial loans is only just getting under way.

* The portfolio manager does not have the same standing in banking as he or she has in the securities world.

* Banks have relationship managers whose objectives may easily conflict with portfolio management goals.

Portfolio management can be employed in banking but with differences from the approach used for securities. Also, there are antitrust pricing issues and proprietary concerns that inhibit banks from cooperating on the development of common tools.

Banking Models

There may not be a standard model for portfolio management in banks, but based on informal discussions among bankers, some published articles and government reports, and a recent Robert Morris Associates survey, there appear to be some common ingredients in various approaches. As in MPT, the first two ingredients are generally rate of return and risk. However, the definitions of these measures differ. Also, banks use risk ranking as well as trial-and-error techniques rather than mean variance analysis to find the best mix.

Rate of Return

The starting point for measuring rate of return generally has been loan spread, although in some cases it is spread or interest yield plus value increases or decreases. The numerator in the return calculation tends to be revenue minus funding costs, operating costs and expenses, and expected cost of credit losses, while the denominator tends to be risk-adjusted or economic capital, not regulatory or accounting capital. Not all models focus on just capital--some focus on outstandings, by adjusting the notional amount to a particular loan equivalency based on, for example, established tenor, collateral, risk-rating and/or capital benchmarks.

Risk

There is an extensive range of risks in banking, so there can be many ways to measure it. Risk can include that associated with rates, liquidity, markets, and legal and regulatory issues. But the prevailing focus in bank models thus far has been on credit risk.

Credit risk can be looked at in terms of two elements: expected credit losses and unexpected credit losses. Expected credit loss is the average loss rate anticipated over time from a credit or from a group of loans of similar quality. Bankers plan and price for "expected credit losses" and can live with those expected credit losses. But this is only the starting point for identifying risk. It is the surprise in "unexpected credit losses" that troubles bankers. They generally think of credit risk as the unexpected credit loss rate, which is the volatility of actual loss rates that will occur around expected losses.