Dear Friends,

This elaboration on Balanced Scorecard will help us know more about the Balanced scorecard. What are its benefits, its limitations and also how to overcome these limitations. We will also discuss why and how performance measures run down, i.e., become ineffective with use. 

Balanced Scorecard : Its limitations and how to overcome them.

Balanced Scorecard(henceforth BSC) was given by Norton and Kaplan. According to them a BSC translates an organisation’s mission into a comprehensive set of performance measures that provide the framework for a strategic measurement and management system.

Norton & Kaplan do not provide any guidelines on how to combine these dissimilar measures in order to appraise the performance of the employees and compensate them.More the number of measures and , more difficult it is to combine them. Therefore BSC is primarily a communication tool and not a tool for mapping performance onto compensation.

Let us now understand why it is so? We know that the performance of a firm cannot be understood well with a single measure. Therefore we need several financial as well as non financial measures. However the challenge lies in selecting the right non financial measures, i.e., the ones which look ahead and help to predict financial performance of the organization. Secondly the greater challenge lies in combining these various dissimilar financial as well as non financial measures in order to appraise and compensate the employee performance along one dimension- that is better or worse. In order to understand the difficulty in this task we will see the case of GFS ( Global financial Services- a US firm).

CASE: Earlier the compensation system at GFS was based only on single earnings measure. Then they implemented BSC. BSC was formula based i.e., explicit weights were attached to various measures in order to assess , appraise and compensate the employee’s overall performance. Later as the employees started gaming the formula ie., ( they achieved high levels of performance on easy targets like sales and ignored the tougher ones like customer satisfaction so that the measure stopped yielding the underlying performance sought) GFS shifted to a subjective compensation system based on BSC. Under the new system weights were assigned to various measures only after measurement and the employees were not clear about who got what and why. This resulted in widespread resistance from the employees. Consequently GFS reverted back to a compensation system based on only sales and earnings.

Problems highlighted by the above case are:

1. Initially when GFS was using formula based BSC, it was gamed by employees.

2. Subjective BSC caused uncertainty. The weights for various performance measures were not explicit. Hence the process of rolling up from performance measures to overall appraisal rating and then compensation package was not clear to the employees.

3. When GFS adopted a subjective BSC, there was an imbalance between weights on financial and non financial measures. Performance on financial measures was rewarded more. Therefore the employees delivered on what was being measured instead of delivering on the underlying performance sought.

It must be noted here that some subjectivity is desirable because it prevents gaming and helps in adjusting the compensation system to the changing business environment. However in GFS’s case the level of subjectivity was too high , so much so that the employees could not understand the connection between performance measurement and evaluation of performance and compensation for this performance. The tension increased all the more because of use of a large number of measures.

Conclusion:

1. Measures , financial ( current performance measure) and non financial ( future performance measure) should predict long term economic performance.

2. We need to combine the non financial measures effectively.

3. The way of combining them should be such that gaming as well as uncertainty can be kept under control.

4. This can be done by adopting measures that are parsimonious, i.e., which can be applied all over the organization. For this we need to distinguish effectively between measures that predict long term economic performance of the firm and those that do not.
Research results in case of GFS:

1. GFS found that instead of a single measure GFS satisfaction measure , specific measures i.e., Branch Quality Index had a greater impact on long term revenues and margins of the firm.

2. Even within this measure i.e., Branch Quality Index general measure like overall satisfaction with the branch which was earlier given a weight of 45% was found to be unimportant . Instead measure ‘ please rate the quality of the teller who served you’ was found to be directly linked directly with the revenues and margins and indirectly with the perceived quality of the branch.

Lessons learnt:

1. It is possible to find Non financial measures that predict financial performance of firms like GFS that have three characteristics:

a. Many branch / business units (less than 30)

b. Similar functions across the various functions.

c. Business units are responsible for their financial performance.

2. Non financial measures predictive of financial performance can be used to appraise and compensate performance provided their limitations are understood i.e.,

a. Measures may and do change with change in business environment—they may be gamed or run down.

b. Measures that gauge the day to day functioning of the organization ( such as the quality of the teller who served you) are more likely to impact the bottomline than generic measures that are far removed from actual functions of the firm.

c. This means that the management needs to arrive at a proper trade off between generic and specific measures.

REMEDY:

Above discussion helps us to come to grips with the problems inherent in using BSC for appraising and rewarding employee performance. Let us now discuss what approach must be adopted in order to make the process more effective and foolproof.

We need to shift to a Balanced Performance Measurement which can be defined as “measuring, appraising and compensating both financial and non financial performance”.

Requirements for Balanced Performance Measurement:

1. Choose non financial measures that look ahead alongside financial measures that look behind.

2. Choose the measures initially in context of a business model or a statement of the firm’s strategy.

3. Consider trade off between generic and specific measures. Generic measures can used for lateral comparisons, they flow top down but are less likely to look ahead. In contrast specific measures originate from bottom, but cannot be compared across units. Composite measures like ( branch quality index) can help to reduce measurement error.

4. validate measures by testing the business model statistically. Ex: Sears started with 70 non financial measures and ended up with 10 efficient ones after statistical measurement. However this is applicable only to firms like Sears which have less than 30 units and not to highly diversified and/or functionally organized firms.

5. Anticipate the measures will change, either due to unforeseen changes in the environment or because of run down. Therefore managers must remember that even the best measures must be checked from time to time.

6. Combining Dissimilar Measures: the management needs to choose between formulaic and subjective combination of measures .

7. Manage distortions caused by compensation formulas gaming can be dealt with by assigning threshold limits to each measure and withholding compensation if any of these limits is not achieved.

Another problem here is inattention to measures whose %age weightings in the compensation formula is small. This can be managed by consolidating measures weighted less than 10%.

8. Combining distortions caused by subjectivity : Subjectivity can result in diminished expectations , lower motivation and therefore lower performance. This can be overcome by explaining to people and making them understand how the performance ratings will be appraised and linked to their compensation. Another problem of using subjective measures is reversion to unbalanced measurement. This can be dealt with by determining statistically whether non financial measures have been factored into appraisals or not and then recalibrating appraisals and bonuses if necessary.

9. Anticipate that the performance measurement system will change.

Here it is important to discuss what is meant by ‘Measures have a tendency to run down with time ‘. Why does it take place? How to deal with this problem?

Let us first discuss the premise of the concept:

1. All performance measures are 2nd best indicators of an uncertain future.

2. Some 2nd best performance measures are better than others.

3. People generally improve what is measured. But sometimes people improve what is measured without improving the performance that is sought.

4. Slowly as people improve on these measures, the range of variance reduces and it becomes difficult to differentiate between good, average and bad performance.

5. This is known as the Use-It-And-Lose-It Principle.

Conclusion: Performance measures are dynamic and need to be changed with time.

Why do Performance measures run down?

1. Positive Learning

2. Perverse Learning

3. Selection

4. Suppression

5. Social Consensus

1. Positive Learning : This reflects learning that occurs as organizations observe one another and converge w.r.t structure and performance. This phenomenon is also known as ‘Organizational Isomorphism’i.e., convergence where similarities have replaced differences (ex: organizational norms) related to permanence and stability.

Convergence in general terms is related to permanence and stability. However in performance measurement , it indicates change. Ex; Gross Mortality Rate in hospitals has come down from 42% to 2.2% , therefore it is no longer possible to distinguish between good and not so good hospitals on this basis and we need to use some new measure.

2. Perverse Learning: This refers to wrong lessons learnt. When reducing variance, i.e., while improving increase in mean occurs without effect on and perhaps to the detriment of actual performance outcomes. Ex: interviewers are sometimes rated on the basis of the number of interviews they conduct. This results in their wrapping up interviews fast without assessing the candidate adequately.

3. Selection : This is an outcome of learning. Ex: Retaining high performers, removing low performers, overtime and other activities reduce the range of variance in performance.

4. Suppression : Differences in performance results are often suppressed especially if these differences persist. Ex: assessment of organizational performance.

5. Social Consensus: Ex: IPOs- Whenever a new IPO is announced, initially its price fluctuates widely. Later as market’s appraisal of the IPO spreads, it trades in a narrower range. It is also known as the ‘seasoning effect’ for IPOs.

Here it must be noted that turbulence in the environment reduces the rate of run down of measures. However in such a scenario, the measurement’s validity itself becomes questionable.

New measures must be weakly correlated or unrelated to the old measures but not strongly related or else the new measures will also run down at the same time. Similarly the new measure should not be negatively correlated to the measure i.e., the new measure must be different but not antagonistic to the old measure.

CONCLUSIONS:

1. Performance measures tend to exhaust themselves as strategies succeed. Ex: GE ( 1950s-60s) used elaborate measures like operational, market and work management measures. In 1970s Jack Welch introduced the measure of firms being either first or second in profit or growth. In 1990s GE shifted to 360 degree feedback and in later 1990s they are using quality control (black belt managers) as a measure.

2. Smaller firms , less confident may stick to spent metrics and consequently fail to make strategic choices and focus on a limited set of objectives.

3. Firms with lowest correlations among Performance measures have highest rates of business growth.

4. Combined forces of learning and selection produce better performance, hence lower variance and therefore little consequence for the game. Ex: the number of defects in cars are now negligible and therefore cannot effect and predict future car sales and profits. Thus new measures are required.

Hope You found the article helpful. Please give me your feedback and experiences on this topic so that I can learn more.

Cheers

Rajesh