CUBenchmarker

Power Benchmarking

Improving your credit union's performance is the most important function of your board and management team. No matter how good your credit union is, or thinks it is, it can always be better. The best credit unions always strive not only to beat the competition, but also to beat themselves. Managers and directors are obligated to members to provide the best possible lending and saving services at the lowest possible cost.

The practice of benchmarking can help you identify areas where your credit union can improve its performance. This is particularly useful for credit unions because credit unions have access to data for other credit unions, and other credit unions are usually willing to share information since they typically do not compete with each other and they belong to a common cooperative movement. Information sharing is central to the benchmarking methodology. When it comes to benchmarking, credit unions have a clear advantage over for-profit banks, which cannot legally work together and share information, even if they wanted to.

Benchmarking done incorrectly, however, can be worse than not doing benchmarking at all. Poorly generated benchmarks can be misleading and move your credit union away from best practices. Management can make the wrong decisions under the cloak of a sophisticated methodology clothed with an array of figures and charts.

The Flaws of Traditional Benchmarking

The typical benchmarking exercise begins by identifying an area for improvement that can be expressed numerically. There are many examples--ROA, member growth, number of mortgage loans, capital to assets, and deposit accounts. The second step is to search a credit union database to identify the best-practice standard for this single variable. The third step is to develop a strategy to achieve (or move toward) best practice.

There are at least three reasons why this procedure is flawed.

1. A single output alone does not measure credit union performance. Your credit union provides services using resources or inputs. For example, a credit union uses managers, other employees, a building, and a data processing system to provide services to members with a level of safety and soundness. It is wrong to benchmark to one output and ignore inputs.

   Consider the problem. The ability to provide services (outputs) depends on access to resources (inputs). Small credit unions with more limited resources cannot provide the same services as large credit unions with less-limited resources. For example, suppose the best value for operating expense over assets among all 10,000 credit unions is Navy Federal Credit Union with $20 billion in assets.

   No matter how large or small a credit union is, it makes no sense to adopt this as a best-practice benchmark. Navy Federal is able to achieve this standard using resources beyond the means of any other credit union. It's neither a fair, interesting, or useful comparison even if Navy Federal is the best. Moreover, even learning and attempting to emulate how Navy Federal operates could be counter-productive because it's a fundamentally different entity. The crude generation of single-variable benchmarks that focuses on one output (or one input) and ignores inputs (or outputs) is a mistake.

2. A best-practice credit union based on a single variable might be an inappropriate benchmark partner because it might not represent superior performance. The problem is that a credit union produces more than one output. Superior performance in one output could be combined with inferior performance in other outputs. This could reflect a different business plan, which is not particularly surprising since the services a credit union provides are determined by the characteristics of its field of membership. It makes little sense to adopt a benchmark partner that could be a worse rather than a better credit union. Single-variable benchmarks that ignore the full specification of inputs and outputs that define the business plan are susceptible to this problem.

3. Performance is a continuum that ranges from the most inferior credit union to the most superior credit union. It's useful to obtain an overall perspective on where a credit union stands. This requires information on worse as well as better credit unions to position the credit union in the performance distribution.

In the quest for a benchmark, there's no particular reason to fixate on the most superior credit union that is dubbed "best practice." In fact, there might be good reasons to adopt a benchmark partner that is superior, but not the very best.

For example, the very best credit union might be an "outlier" due to idiosyncratic circumstances or errors in the data. A less than best practice benchmark might be appropriate as a more realistic target or an interim goal. A more reliable benchmark might come from a dense area of the frequency distribution where it's evident that a number of credit unions are able to achieve the standard.

Although single variable benchmarks don't represent a best practice in benchmarking, they can provide some insights if used cautiously. They do have the advantage of being a single number, they're easy to interpret, and easy to calculate. Rather than generate a single best-practice value, however, it's much more valuable to generate a histogram that includes a full set of inferior and superior credit unions.

The two histograms of asset growth in Figure 1 illustrate this enhanced approach to single variable benchmarking. Suppose that one objective is to evaluate the credit union's performance in achieving asset growth. Is the top histogram or the bottom histogram the better credit union? Each might have a similar number of superior and inferior credit unions. Each might have the same best practice credit union. Notice that the in the top histogram, however, the superior credit unions tend to be close to credit union A--they're not superior by very much. The inferior credit unions tend to far from credit union A--they're inferior by a lot.

The bottom histogram exhibits the opposite pattern. Credit union A (the top histogram) is performing better than credit union B. To obtain an overall evaluation of performance, it's necessary to interpret a credit union's position relative to the performance of better and worse credit unions.

Suppose that a second objective is to identify a benchmark for asset growth. Rather than presume that the obvious standard is the credit union that has achieved the highest asset growth, the enhanced approach to benchmarking recommends that less than best practice credit unions be considered. In the second histogram in Figure 1, there might be good reasons to select a benchmark from credit unions in the highest bar toward the middle of the distribution of superior credit unions. Compared to the bar that corresponds to best practice, this benchmark might be a more realistic choice, and the fact that a large number of credit unions are able to achieve this standard gives it added credibility.

In summary, single variable benchmarks do have some advantages and they can be useful if used cautiously. The enhanced approach to single variable benchmarks focuses on the distribution of better and worse credit unions. This framework provides an assessment of performance in terms of the single variable and a basis for selecting a benchmark goal to improve performance.

Power Benchmarking

Power Benchmarking is a technique to generate a numerical benchmark for a credit union's overall business plan. Rather than focusing on a single variable such as asset growth, Power Benchmarking generates a benchmark based on the resources a credit union has to work with and the services it offers members. Power Benchmarking has the following advantages:

• A distribution of single numerical benchmarks is calculated analogous to the histograms discussed earlier.

• Benchmarks incorporate your credit union's overall business plan.

• Benchmarks take the individual nature of your credit union into account. They're customized to a credit union's operating environment and its mission.

The first step is to characterize a credit union in terms of its resources and the services or outputs that it offers members. This quantifies the mission or the business plan. A credit union might be interested in alternative specifications depending on the objective of the analysis. Figures 2 and 3 illustrate two of many possibilities.

Figure 2 focuses on financial performance. The resources are total assets. The outputs are net income, asset growth, and capital to assets. The cloud over the top represents controls. Benchmarks are generated based on credit unions with a similar number of branches, a similar ratio of mortgage loans to total loans, and a community charter.

Figure 3 focuses on providing services to members. Resources are total assets, income from capital, and fee income. Service outputs are quantity of loans, interest rate on loans (lower is better), quantity of deposit accounts, interest rate on deposit accounts (higher is better), variety of services, and the number of transactions (measured by the number of share draft accounts). The controls are the same as in Figure 2.

The objective is to benchmark a credit union's performance in terms of its multiple input (resources)/multiple output (services) business plan and to generate a single numerical value. The Peer Group Family Tree for Credit Union A in Figure 4 illustrates the approach. Credit Unions C, D, F, and G are high performing peers for Credit Union A.

These credit unions outperform Credit Union A in terms of its business plan, and they're able to achieve this using the same or fewer resources. Credit Unions E and I are low performing peers for Credit Union A. These credit unions underperform Credit Union A in terms of its business plan, and they use the same or more resources than Credit Union A. Measures of distance are required to convert the high and low performing peers into single-number benchmarks.

High Performing Peer

A credit union is a high performing peer if it produces more of all outputs with the same or fewer inputs. Consider the meaning of a high performing peer (HPP) in the context of the service-to-member business plan illustrated in Figure 3. A credit union is a HPP for Credit Union A if it provides more loans at a lower interest rate, more deposit accounts at a higher interest rate, greater variety of services, more transactions, and has a lower operating expense. Consider the significance of a HPP:

• HPPs take the outputs into account and the inputs or resources available to the credit union. A credit union that produces more with more resources isn't a relevant peer and fails the HPP test. Navy Federal Credit Union would not be a HPP for almost all (if not all) credit unions no matter how well it performs because it has far more resources to work with. Contrast this to the traditional benchmarking approach that identifies best practice without taking into account either multiple outputs or resources.

• HPPs are customized to a credit union's mission or business plan. The input-output specification quantifies the mission. In this case, it's a service-to-member focus. HPPs use fewer resources to better accomplish the objectives of Credit Union A. HPPs are doing more with less. These are very relevant and appropriate peers for benchmarking.

• There isn't a single best-practice HPP, but rather a number of HPPs. It will be shown that these HPPs can be ranked and represented in terms of a histogram.

• The list of HPPs can be screened to include those that have the same or fewer branches, the same or a higher mortgage-to-total-loan ratio, and a community charter.

Low Peforming Peer (LPP)

A credit union is a low-performing peer if it produces fewer of all outputs with the same or more inputs. Consider the meaning of a LPP in the context of the service-to-member business plan illustrated in Figure 3. A credit union is a LPP for Credit Union A if it provides fewer loans at a higher interest rate, fewer deposit accounts at a lower interest rate, less variety of services, fewer transactions, and has a higher operating expense. Consider the significance of LPPs:

• LPPs are following the same mission or business plan as Credit Union A, but Credit Union A accomplishes it better with fewer resources. LPPs are customized to Credit Union A's mission and circumstances.

• LPPs are just as important as HPPs. Obviously, there's no reason to benchmark to an LPP, but there are good reasons to know where your credit union stands in terms of inferior credit unions that it dominates as will as superior credit unions that dominate it.

Measure of Performance

A good measure of performance includes HPPs, LPPs, and a measure of distance. The distance measure captures the fact that some HPPs are more superior than other HPPs, and likewise for LPPs.

Figure 5 illustrates the measure of distance for a simple specification where loans and deposits are the only outputs, and Credit Unions B, C and D use fewer resources than Credit Union A. Credit Unions B, C, and D are all HPPs for Credit Union A, but they're not necessarily equivalent. Credit Union B provides twice as many loans and twice as many deposits as Credit Union A, and it has a distance measure equal to 2. Credit Union C provides 1.5 times as many loans and deposits as Credit Union A, and it has a distance measure equal to 1.5. Credit Union D provides 1.5 times as many loans and twice as many deposits as Credit Union A, and it has a distance measure equal to 1.5. When the performance differential doesn't come out evenly across outputs, the distance measure is the smaller differential that captures the maximum equiproportionate expansion that Credit Union A could achieve relative to Credit Union D in this case.

Figure 6 illustrates the analogous procedure for LPPs. Credit Unions E, F, and G use more resources than Credit Union A and produce fewer loans and deposits. They're LPPs for Credit Union A. Credit Union E, for example, produces 0.75 of the loans of Credit Union A and 0.75 of the deposits of Credit Union A, and it has a distance measure of 0.75. The distance measures for Credit Unions F and G are apparent based on the previous discussion.

Power Benchmarks

Figure 7 illustrates the LPPs and HPPs for Credit Union A according to their distance measures, and Figure 8 illustrates a sample histogram of LPPs and HPPs using the distance measures. These just happen to be the same histograms as were used to illustrate single-variable benchmarks. They are merely illustrative.

Consider the significance of the Power Benchmark histograms:

• The flaws of single variable benchmarks have been corrected. Power Benchmarks are based on the full specification of the credit union business plan that characterizes its mission. Both inputs and outputs are incorporated into the benchmarks.

• Power Benchmarks are a single number (the distance measure) and have a clear and convenient interpretation. For example, suppose a potential benchmark partner has a score of 2.5. This means the HPP can produce at least 2.5 times as much of each output with the same or fewer resources.

• The histogram of LPPs and HPPs provides a visual representation of where a credit union stands in terms of accomplishing its business plan compared to other credit unions with a similar set of objectives. This can be interpreted in the same way as for single variable benchmarks discussed earlier.

• The power benchmark histogram can be used to select an appropriate benchmark partner, and that benchmark partner need not be the one furthest to the right.

Questions and Answers

1. How do I appropriately specify inputs and outputs that capture my business plan?

   This is an opportunity to carefully think about exactly what you intend to accomplish. However, it is not necessary to settle on a single specification. A full evaluation of your credit union would include the generation of Power Benchmark histograms for a number of different input-output specifications, each with a different focus.

2. Once I select a business plan, generate the histogram, and select a bar that represents a reasonable target, then what do I do?

   Identify the credit unions that inhabit the bar. The Power Benchmarking approach has essentially served as a preliminary screen. Now you can select a specific credit union that appears to be a particularly good match to your credit union. Compare the benchmark credit union's values for inputs and outputs. These are specific benchmarking targets.

   Next, you can examine the detailed financials of the benchmark credit union, or even better--call up the CEO to arrange for a detailed discussion to learn what the benchmark credit union is doing that could improve your performance.

3. What are some other uses for Power Benchmarking?

   Return to the CUBenchmarker
   start page.
   You can use the power benchmarking approach to benchmark growth, benchmark the impact of a potential merger, and to benchmark alternative strategic plans.