| Asset-Liability Management |
We were once shown a faxed piece of bank-directed research that started with something like, "Every portfolio is like a snowflake. Each bank has different needs, and so there is a bond for every portfolio and a portfolio for every bond." This kind of statement leads to other conclusions, like, "Since every portfolio is unique, then why compare their performance?" But the reality of bank portfolio management, as we have seen over the years, is that bank portfolios are very much more alike than they are dissimilar. Chart A shows just how tightly banks were gathered in their year-ending third quarter 1996 total returns (latest full year of data available).
Similar graphs show tight variances for 1994 and 1995. Generally speaking, a group of "unique" portfolios should have a much wider variance. Furthermore, when one looks at portfolio performance according to size or community type, one finds no direct correlation. But let's give a better reason why banks can and should compare their portfolio results.
NEW TOOL NEEDED
While we have seen many banks with very, very different loan portfolios across the country, bankers seem to be comfortable (at least when the comparison is favorable!) comparing their loan performance to others'. And while liabilities range from totally interest-rate sensitive to insensitive, bankers do not generally object to comparison in this area, either. But for some reason, in the third major financial instrument within the bank (the securities portfolio) there has been no tradition of comparison. This is very strange when one considers that the investment portfolio is the one area of bank management where everyone plays on the same field. All banks have access to the same securities! Then why don't we compare results? The answer is that we need a good tool to do so.
The most obvious solution would be to use call report data that measures both income and market values in the investment portfolio. But there are some problems with this methodology. First there is the lag time for the information. Call report data is usually not available from data providers like Sheshunoff, Plansmith, (F.I.R.E.) or IDC Financial Publishing for four or more months after quarter's end. The portfolio manager who wonders, "How has my portfolio performed over the last quarter?" cannot get his question answered for up to six months.
Second, these data can significantly distort short term results due to their lack of effectively dealing with the timings of cashflows into and out of a portfolio. IDC Financial publishing tries to deal with this issue by averaging quarterly portfolio assets to calculate its return numbers. But if a portfolio is restructured or significantly added to or reduced in size near quarter's end, the results can be under or over-stated. If the portfolio undergoes significant change early in a quarter, results can be similarly misleading.
So if call report data, at least as it is currently compiled, is found lacking, what would the best benchmark look like? Obviously it should seek to replicate a typical level of both risk and reward for a typical bank portfolio. The number should be available much sooner than traditional call report data and it should be subject to consistent and higher quality (bid side) pricing than many portfolios and their accounting systems report. But should the benchmark be a yield benchmark, or a market value benchmark or a total return benchmark? Regular readers of this column can surely guess what we would argue for!
As we wrote in our last column, total return ultimately flows through to earnings and is the best measurement for long-term shareholder value. Total return for a given period records income plus or minus gain or loss on a security. As we have written many times, market value represents the present value of your security's future cashflows versus other investor's securities purchased at today's yield (discount rate). Ultimately a portfolio manager who earns total returns consistently higher than average will out-income his peers. Total return looks past a single-year statistic and gives the shareholder a real look at how the portfolio has contributed to the institution's wealth, not just it's income.
In our tradition of reporting the realities of modern-day bank investment portfolio management, Betzold, Berg & Nussbaum, Inc. is pleased to announce that it will be publishing a new bank-specific bond index, monthly, in ABA Banking Journal.
HOW THE NEW INDEX WORKS
The Betzold Berg Investment Portfolio Index is designed to realistically track a typical bank portfolio composed of the same percentage of sectors as the average bank. The portfolio consists of six sectors: Treasuries, munis, agencies, fixed-rate mortgage-backed securities and/or CMO/REMICs, adjustable-rate securities, and "other" securities as defined in the call report data. The portfolio will be rebalanced monthly as principle pays down and to the latest available call report data on portfolio allocation. In order to provide the most useful information to the greatest number of banks and to eliminate statistical "noise," we eliminate all portfolios below $10 million and above $1 billion. This represents the average allocations of more than 7,000 banks.
The current index is weighted identically to the last call report data from end of third quarter, 1996. Those holdings are comprised of 24.52% Treasuries, 12.26% municipal bonds, 24.38% agencies, 19.93% fixed-rate mortgage or mortgage related products, 6.09% "other" securities, and 13% adjustable-rate securities. While fed funds, other interest bearing deposits and off-balance sheet items may be included in the average portfolio manager's duties, this index is designed to measure the portfolio manager's skill in the fixed-income cash market.
It is our belief that many portfolio managers who use off-balance-sheet products, would serve their shareholders better if they focused on the risk/reward characteristics of their cash portfolios, before they sought to hedge their bets. We expect many to adjust their portfolio returns (by off-balance sheet results) when comparing their performance to the index, but we believe the index should be as pure as possible in its measurement of cash fixed-income investing. As the AICPA considers revisions in accounting for off-balance-sheet transactions, we will reconsider the effective measurement of such portfolio activities.
WHAT THE NUMBERS SAY
The information in Chart B projects the past performance of the average portfolio as calculated by Betzold, Berg & Nussbaum, Inc., based on the latest available call report data returns and allocations. "Average" in this context refers to the median portfolio returns of banks with more than $10 million and up to $1 billion in portfolio assets.
The index number tells you, for the latest month prior to publication, how a typical bank portfolio has fared. Our ultimate goal is to come as close as possible to tracking the average portfolio on a real-time basis.
As you can see from the study we did in creating the index, that applying this methodology over the last three years would have produced very accurate numbers (Chart C). While our study shows accuracy in tracking of 10 basis points or less, we will be quite pleased if our historical accuracy is within 50 basis points on a monthly basis, with an average of being within 30 basis points annually. Obviously the shorter the time horizon, the more price volatility will affect the results.
While the monthly number should give you a good feel for the price movements of the typical portfolio, the longer term numbers, one year to three years, are a fairer way to measure real performance. As the index grows in longevity, we will report numbers for the last month, year-to-date, last year, last two years and last three years. Initially, as here, we will supplement the index numbers with historical call report data to give a longer term picture of the average portfolio.
 |
 |