One of the most profound changes in business is the new emphasis on performance. The cover of the January 3, 1994 Forbes magazine showed an executive with a cane around his neck, about to be pulled offstage. The headline read: "Perform - or else".
Urgency about financial performance is accompanied by heightened awareness that financial performance cannot be sustained unless the non-financial underpinnings of financial performance - innovation, productivity, product quality, customer service, customer satisfaction - are measured and improved.
This message is at the core of the quality movement and reverberates in countless articles and the nearly 400 books published each year on quality management. Corporate re-engineering efforts communicate a similar message: financial performance depends on inventing processes that are both more efficient and serve the customer better - which means that current processes must be understood, measured and improved.
More From This Section
Many businesses are grappling with the choice of non-financial measures to complement the financial measures they now have in place. Few, however, are comfortable with the choices they have made and some have given up.
"It's impossible," more than one manager has told me.
This brings to mind a quip made some years ago by the psychoanalyst Ralph Greenson: "Not only is psychoanalysis impossible, but it is also very difficult."
The difficulty in this instance is not measuring non-financial performance. Measuring is easy. Today, almost any aspect of non-financial performance can be measured and monitored inexpensively and, often, unobtrusively.
One pharmaceutical house tracks the activity of its computer support staff by counting e-mail messages between line businesses and the support staff. Customer satisfaction surveys are legion. The difficulty is not measuring but, rather, knowing what to measure and knowing what requires attention and what can be ignored - at least for the time being.
Since all aspects of non-financial performance potentially affect long-term financial performance, managers are tempted to measure everything about people, processes, products and customers that the quality gurus and re-engineering experts believe important. The temptation to measure is especially powerful when the competition is engaged in a frenzy of measurement - how can we forego the potential advantage of measuring and benchmarking our results against theirs?
This temptation can be overcome by remembering that to measure everything is ultimately to create confusion rather than information. Physicists have long understood that a quantum world does not permit both the position and momentum of a particle to be known. In his new book Fire in the Mind, George Johnson generalises this quantum principle to information: "If you know everything, you know nothing."
Finding an informative and manageable set of financial and non-financial measures for an organisation, then, is one of the principal challenges facing managers. My research suggests that some firms are able to do this, but only when they have understood the key differences between financial and non-financial measures, how their measures should be configured (I will explain this presently) and some of the organisational conditions contributing to effective measurement.
Differences between financial and non-financial measures
One difference between financial and non-financial measures is their sheer number and relatedness. There are many more non-financial than financial performance measures. There are relatively few financial measures because most of them, financial ratios especially, are governed by accounting conventions and are standardised.
Many financial measures, moreover, are closely related because they are subsets or supersets of other financial measures. EBITDA, for example, is earnings before interest, taxes, depreciation and amortisation. Cash flow is EBITDA plus additions to working and fixed capital.
Innovative financial measures do appear from time to time, but whether they are truly new and different is debatable. A recent addition to the stock of financial measures is EVA, or economic value added, a trademark of Stern Stewart & Company. EVA is roughly the difference between a company's total returns to bondholders and shareholders and returns that could have been earned from investing in other companies at similar levels of risk. EVA, in other words, is returns less the economic cost of debt and equity capital - which means that a company can be profitable but under water by the EVA measure.
Non-financial measures, unlike financial measures, are not governed by accounting conventions and are rarely standardised - indeed, near anarchy prevails in the realm of non-financial measurement. Without standardisation, non-financial measures have multiplied rapidly. The only measures gauging process and product quality 30 years ago were rejects and returns - remember the amount of rework once done in automobile manufacturing.
Today, there are many measures of process and product quality: speed, conformance to product specifications, conformance to fixed quality standards such as ISO 9000, meeting or exceeding competitors' process and product standards.
Thirty years ago, hardly anyone had heard of customer satisfaction. Companies looked to market size and market share statistics to gauge whether or not they had met customers' needs. Today, almost all large companies survey customer satisfaction, although the exact measures differ. And as the evidence showing a link between customer satisfaction and profitability mounts, companies are beginning to look to measures of competitive customer satisfaction - are our customers more or less satisfied than our competitors' customers?
Aside from their number, there is less relatedness among non-financial performance measures than among financial measures.
Conformance quality - conformance to specifications - is an attribute of products and services, whereas customer satisfaction is an attribute of customers. Conformance quality may contribute to customer satisfaction (other things being equal, people prefer products to meet specifications and hence their expectations) but it also may not (people do not appreciate the cost of over-engineered products). There is no necessary connection between the two.
Not only is the relatedness of non-financial measures weaker than the relatedness of financial measures, but the relatedness of a non-financial measure to financial performance can be determined only through statistical evidence that can take months or years to gather.
Product quality, speed and convenience, or satisfaction with the manner in which the transaction was conducted may affect customers' willingness to engage in further transactions and hence future financial performance depending on circumstances - which can be complicated.
Another key difference between financial and non-financial measures is the tendency of non-financial measures to run down with use, to lose variance and hence the capacity to discriminate good from bad performance. The classic case of running down is batting averages in major league baseball in the US.
Differences in batting averages between the best and the worst players have diminished substantially since baseball statistics were first kept in the 1870s. There are no more 400 hitters.
Several explanations for diminished variation in batting averages come to mind. Perhaps the most cogent is suggested by the palaeontologist Stephen J. Gould, who argues that improvements in player selection due to the creation of the minor league "farm system" coupled with improved coaching and training of major league pitchers and batters have caused differences in batting averages to shrink.
I have found that many performance measures exhibit diminished variance over time, for example, occupancy and length-of-stay statistics for hospitals, safety statistics for nuclear power plant, yields of money market mutual funds and even customer satisfaction - several companies report that customer satisfaction now exceeds 90 per cent and is not expected to improve further.
I have also found that organisations seek new and somewhat different performance measures when existing measures lose variance. Batting averages hardly figure in baseball contract negotiations nowadays - slugging percentage, on-base percentage, run production and the like have taken the place of batting average.
Hospitals have moved away from functional performance measures such as occupancy to disease-specific morbidity and mortality rates. The Nuclear Regulatory Commission is continually searching for new safety measures. Money market mutual funds compete today on service as much as on yield. Some companies are introducing measures of customer loyalty to complement customer satisfaction.
Although diminished variance in measured performance often occurs due to improvement, perverse learning, or gaming, also causes variance to diminish. Teachers teach to test. Police investigators elicit multiple confessions from suspects to maintain clearance rates. Workers learn how to meet their "bogeys" exactly, sometimes by hiding or suppressing output, sometimes by sharing output with less productive colleagues.
The problem for managers is knowing when diminished variance signals improvement and when it signals gaming or outright deception. Usually, managers do not even try to distinguish improvement from gaming and deception. Instead, they search for new measures where variation exists and cannot be gamed away immediately.
Given the differences between financial and non-financial measures, why bother to measure non-financial performance at all?
The answer is that financial measures summarise past performance well but predict future performance poorly - very poorly indeed. Consider the ultimate financial performance measure - share price.
Shares of companies singled out as poor performers by the Council of Institutional Investors from 1991 through 1993 have since outperformed the market by about 50 per cent. The 20 highest-performing mutual funds in each year from 1982 to 1992 ranked, on average, at about the middle of the pack in each subsequent year.
Outstanding financial performance, to be sure, confers some advantages on companies. Those with outstanding financial results enjoy better access to debt and equity markets and better reputations than their less outstanding counterparts. But as often as not, outstanding financial performance is followed by mediocre performance.
It is no wonder, then, that many managers regard financial measures as backward-looking, as rear view measures telling you how well they did but not how well they are likely to do, and seek non-financial measures capable of predicting financial performance.
How to configure measures
I have some rules of thumb for combining non-financial and financial performance measures. The rules are simple in principle, although somewhat more difficult to implement in practice.
The number of measures: there should be more than one or two measures gauging progress toward strategic objectives (multiple constraints will help to keep gaming in check) but fewer than five or six (remember George Johnson's law: "If you know everything, you know nothing").
The balance of financial and non-financial measures: there should be some of both. Including non-financial measures is especially critical, since these are the leading indicators of performance.
The properties of non-financial measures: non-financial measures must meet three requirements. There must be variance or room for improvement - measures that cannot be improved cannot contribute to financial performance. They must be under your control so that you can take actions to improve them. And there must be a clear path from non-financial performance to financial results so that improvement in the former produces improvement in the latter. The first two requirements for non-financial measures - improvable and controllable - are fairly easy to meet. But the third requirement - a clear path from non-financial to financial performance - is more difficult to satisfy because it takes considerable evidence to demonstrate that non-financial performance produces financial results.
The level of constraint among measures: a moderate level of constraint is desirable. Measures should be sufficiently constrained so that gains in any one reflect true performance gains. Constraints should not be so severe that gains in one measure can be achieved only at the expense of other measures. Nor should constraints be so relaxed that gains in any one measure automatically produce gains in the others.
The availability of alternative measures: new measures should be available should existing measures run down.
These rules of thumb were used in configuring performance measures for the distribution organisation of a global electronics manufacturer. The key financial measure for the larger organisation was ROA, return on assets.
Since the distribution organisation was a cost centre rather than a profit centre, its contribution to the numerator of ROA was measured by order fulfilment costs - the lower the order fulfilment costs, the better the returns. The distribution organisation's contribution to the denominator of ROA was measured by the ratio of total revenues to inventory - the higher this ratio, the smaller the fraction of assets tied up in inventory.
Two non-financial measures were chosen as well: the time needed to fill orders and the proportion of orders filled directly from inventory - experience had taught this organisation that delays caused customers to defect to other sources.
Altogether, then, the distribution organisation used four performance measures, two financial measures (order fulfilment costs and the ratio of revenues to inventory) and two non-financial measures (time to order fulfilment and orders fulfilled from inventory). The non-financial measures were improvable, controllable and predictive of financial performance. Together, these measures constrained one another moderately.
To illustrate some of the constraints: time to order fulfilment could not be managed by increasing order fulfilment costs (for example, by using air rather than surface transportation). And the proportion of orders filled from inventory could not be managed by increasing inventories and hence the ratio of inventories to revenue. Better ways of improving performance were required and, in fact, were ultimately discovered.
Moreover, even as these performance measures were being put into place, their replacements were under active consideration by the distribution organisation.
Organisational conditions contributing to effective performance measurement
Effective performance measurement requires centralised control of performance measures. Since the 1950s, most large corporations have operated on an extremely decentralised basis: financial objectives are set for business units and business unit managers are permitted to pursue these objectives as they see fit.
These practices may change as companies begin introducing non-financial measures alongside financial ones and thinking about their overall configuration of measures.
Greater centralisation of companies, I believe, will occur. Centralisation will be needed to standardise non-financial measures, to compare and reward both financial and non-financial results, and to review and replace performance measures.
The distribution organisation of the global electronics manufacturer illustrates the trend toward centralisation.
Its performance measures were decided by the head of organisation and his controller, and the controller retains responsibility for reviewing measures. In another global manufacturing firm, one vice-president has responsibility for performance measures worldwide.
In a global financial services company that historically has operated on a highly decentralised basis, common measures for front-office processes will somewhat erode the autonomy of country-based business units.
These cases, I believe, are not idiosyncratic. They reflect a simple logic of measurement: so long as performance is gauged by a single measure that can be compared readily across businesses, little central control is needed. Once, however, multiple measures, financial and non-financial measures, and the properties of these measures as well as constraints among them must be considered, some centralisation is inevitable.
Summary
There is a growing realisation that financial performance cannot be sustained unless the non-financial underpinnings are measured as well. Financial measures may summarise past performance but they predict future performance poorly. The problem is not the measurement as such, rather knowing what to measure, what requires attention and what can be ignored. Occasionally new financial measures like EVA come along - but there are many more non-financial ones. There is less relatedness among them, conformance quality for example being an attribute of product and services, customer satisfaction an attribute of customers. The statistics on which they are based can take months or years to gather and they have a tendency to "run down" with use. Organisations often seek new performance measures when the old ones lose variance. Progress towards strategic objectives should be gauged by more than one or two measures, but less than 5 or 6. Make sure some of them are non-financial, though there must be a clear path to financial performance. Effective performance measurement requires centralised control of the measures.
Signpost
International Macroeconomy
The Module deals with corporate performance. It will continue in Parts 17 and 19 and topics will include global alliances and regionalisation. The Module began in Part 12 with sections covering "market-driven" companies and competitive societies.
Marshall W. Meyer
Marshall W. Meyer is professor of management and Anheuser-Busch term professor in the Wharton School and professor of sociology at the University of Pennsylvania. He is president of Research Committee 17, Sociology of Organizations, of the International Sociological Association and is associate editor of Administrative Science Quarterly