In what way can the use of ROI as a performance measure for investment centers lead to bad decisions How does the residual income approach overcome this problem?

• Tweet

  Inhaltsverzeichnis Show

  • The Du Pont System
  • Distinctive elements
  • Alleged advantages
  • Inherent ROI Limitations
  • Technical drawbacks
  • Correcting the deficiencies
  • Implementation constraints
  • Developing New Systems
  • Methods available now
  • In what way can the use of ROI as a performance measure for investment centers?
  • In what way can the use of ROI as a performance measures for investment Centres lead bad decisions How does the residual income approach overcome this problem?
  • What is residual income and how it is applied in evaluating new investment project when compared to ROI?
  • Why does the balanced scorecard include financial performance measures as well as measures of how well internal processes are doing?

• Post

• Share

• Save

• Get PDF

• Buy Copies

• Print

The use of some form of return on investment (ROI) as a management control device in evaluating the profit performance of division managers has been widely adopted in many decentralized companies. Yet the evidence shows that this control system has serious limitations, which result from the inability to use ROI to make correct evaluations. The author notes that any criticism of the use of ROI is met with the response, “I agree it is not perfect, but it is the best system available.” This, he says, is the crux of the problem. He believes there are better systems available and offers specific actions that should improve the management control system of any company using ROI measurement.

Nearly every major decentralized company in the United States today uses some adaptation of return on investment for measuring division performance. All these systems, in some respects, encourage division managers to take actions contrary to the overall interests of the company. This occurs because the measurement devices used in these ROI control systems affect division performance inconsistently with overall company performance. Thus an action by a division manager that reduces company profits may show a favorable increase in division profits. In addition to the problems that result from inconsistency between division objectives and company goals, many of these systems have serious limitations as a means of performance evaluation.

The purpose of this article is to describe the ROI system of decentralized financial control, to explain why it creates conflicts between company and division interests, and to describe why it is often of limited use in evaluating division performance.

As a matter of historical note, the ROI approach to financial control was originally pioneered by Du Pont for its decentralized operations, and today in many companies that use some form of return-on-investment measurement it is often referred to as the “Du Pont system.” Nothing in this article, however, should be construed as criticism of the Du Pont company.

I have used the original form of the Du Pont system as the basis for my article because: (a) it was the first major comprehensive decentralized financial control system developed in the United States; (b) almost every major decentralized company today uses some adaptation of it; and (c) the system itself is extremely well documented. At the time Du Pont first developed its return-on-investment control system, it was far superior to anything else then in existence. Over the years, however, flaws have been discovered in the system. Moreover, changes in accounting techniques have created problems that did not exist when the system was first developed. Finally, return on investment is only one of Du Pont’s control mechanisms; yet many companies have adopted the ROI system while ignoring the other elements of control that have made this approach effective at Du Pont. All of the conclusions presented herein, therefore, will apply to systems that measure profit performance in terms of “return on investment.”

The Du Pont System

William T. Jerome has described Du Pont’s decentralized financial control system in this way:

“As practiced by Du Pont executives, return on investment has become a symbol for the company’s system of management control. As such, return on investment is a way of approaching the problem of top-level control rather than a magic formula for solving these problems.”1

Distinctive elements

The chief characteristic of such a management control system is the measurement of financial performance in terms of return on investment. This is a fraction, the numerator of which is the accounted profit earned by the division, while the denominator is the investment assigned to the division. The investment consists of two parts: fixed assets and working capital (inventories, receivables, and cash). The fixed assets are included in the investment base at gross book value (original cost); inventories and receivables are included in the investment base at their actual accounting balances; and cash is an allocated amount (since cash is controlled centrally) that is equal to one-and-a-half times the department’s forecasted average monthly cost of sales less the depreciation taken by the department.

A second important characteristic of this control system is the wide degree of latitude allowed the industrial departments in meeting their ROI goals. “The job of the operating department is thus to make a projected return on investment, not to live within a budget”.2 In other words, there is little concern as to how the individual manager meets his ROI objective as long as he does. For example, he may increase his return on investment in the short run by reducing costs, by increasing sales volume or prices, or by reducing his investment base. Under this system, the important consideration is the impact on the ROI.

A third characteristic is that performance is evaluated quarterly by comparing actual return on investment with the ROI objective and by analyzing the difference. Individual analyses are made on the basis of charts, of which the “formula chart” shown in Exhibit I is probably the best known.

Exhibit I Typical Formula Chart Showing Relationship of Factors Affecting ROI

Alleged advantages

As a measure of division performance, the use of ROI is alleged to have certain advantages:

1. It is a single comprehensive figure influenced by everything that has happened which affects the financial status of a division.

2. It measures how well the division manager uses the property of the company to generate profits. Consequently, a manager will utilize the property at his disposal to its fullest and will acquire additional assets only when they will improve his investment return. It is also a means for checking automatically on the accuracy of capital investment proposals. If an approved project earns less than that shown in a capital investment proposal, the division rate of return will be affected adversely.

3. Rate of return is a common denominator that can be compared directly among divisions only, among divisions and outside companies, or among divisions and alternative investments of funds.

In short, the system is supposed to result in each division manager optimizing his investment return which will in turn result in an optimum total company return. Since the manager is evaluated on his ability to optimize ROI, he will be motivated to do so.

Inherent ROI Limitations

Despite the fact that some adaptation of ROI is a widely used tool in management control in the United States, I feel that a strong case can be made against it because the ROI system has certain inherent limitations in it. For purposes of this discussion, I find it useful to separate these limitations into two types: “technical” and “implementation.” The first type are those conditions which cause incongruities between divisional objectives and company goals, and which result in motivating division managers to take uneconomic actions. The second type includes those conditions that result from the inability, under many circumstances, to evaluate accurately the profit performance of division managers. Let us look more closely at each of these limitations.

Technical drawbacks

The single most important limitation in this category results from the fact that ROI oversimplifies a very complex decision-making process. The use of a single ratio to measure division performance reduces investment decision making to a simple but unrealistic economic model. Under this system, any change in the investment base can be traded off against a specific amount of profit which is determined by the division’s objective rate of return. Consider:

• If a division’s objective rate of return is 10% after taxes, a reduction of one dollar in the investment base has the same impact on performance as a 20-cent saving in costs or a 20-cent increase in profits from increased revenue (assuming an income tax rate of 50%).

• Conversely, an increase of one dollar in any asset will have to provide an increase of at least 20 cents in profits before taxes, or the rate of return will be affected adversely.

Under the ROI control system, the economic trade-off (i.e., the ratio of investment to profits) is constant throughout the division; it is the same (a) for all assets, (b) at all times (at least, until the objective is changed), and (c) for adding additional investments as for reducing the value of investments currently on the books.

Furthermore, although the trade-off between investment and profit is constant throughout the division, it will differ among divisions where their profit objectives differ. This means that inventories in one division will have a different carrying charge from identical types of inventories in another division with a different profit objective.

Since each division is expected to earn an ROI objective, a division manager will not be likely to propose a capital investment unless it is expected to earn a return at least as high as his objective. Thus a division with an objective of, say, 20% would not want to invest at less than this rate, while a division with an objective of 5% would benefit from anything over this rate. Since it is probable that the profit objectives of most divisions are different from the company’s cutoff rate for capital expenditures, this situation can cause inconsistent capital investment actions.

Another complication under the ROI system is that different types of fixed assets must earn the same return. For example, a general-purpose warehouse that can be leased fairly inexpensively must earn the same return as special-purpose equipment that may be subject to considerable potential obsolescence. This means that divisions with high profit objectives will maximize their investment returns by leasing as much of their assets as possible. Conversely, divisions with low profit objectives might improve their returns by purchasing the same type of equipment that the other divisions are leasing.

Use of book value:

A study by John J. Mauriel and Robert N. Anthony indicated that 93% of the responding companies measuring division performance on the basis of ROI included fixed assets in their investment base at either gross book value (18.5%) or net book value (73.2%), or a combination of the two (1.6%).3 To me, though, the use of book value (either gross or net) is unsatisfactory in a financial control system, and I shall explain why.

If gross book value is used, it is possible for a manager to increase his investment return by scrapping perfectly useful assets that are not contributing profits equal to the division’s objective. (When an asset is scrapped, the fixed-assets account is reduced by the original cost of the asset.) If composite depreciation is used (and most large companies use composite depreciation), the division will not even show a capital loss from disposition. This occurs because, under composite depreciation, assets are considered to be fully depreciated when they are retired.

The theory is that the depreciation rates are based on the average life of a group of assets. Some will last longer than the average; others, shorter. No gain or loss is taken, therefore, when an individual asset is retired. Even if unit depreciation is used, the loss will be less than the reduction in the investment base where assets have been depreciated. This means that the impact of scrapping a fixed asset can be favorable to the division’s ROI even where such action is clearly detrimental to the company.

In addition to encouraging the disposition of perfectly good assets, the use of gross book value will have an inconsistent effect on the investment base when equipment is replaced. The gross book value will increase only by the difference between the cost of the new equipment and the original cost of the old. The investment to the company, however, is equal to the cost of the new equipment minus the salvage value of the old equipment. In most cases, the salvage value will be far below the original cost. Thus replacement investments that are far below the company’s cutoff rate could improve the division’s rate of return.

If fixed assets are included in the investment base at net book value (gross book value minus accumulated depreciation), a division manager can reduce his investment base only by the undepreciated value. (When an asset is scrapped, the fixed asset account is reduced by the original cost of the asset; the depreciation account is reduced by the amount of accumulated depreciation applicable to the asset; consequently, the reduction in net book value is the difference between these two amounts.) If the company uses composite depreciation, the investment base will not change because, as stated earlier, all retired assets are assumed to be fully depreciated under this method. If unit depreciation is used and fixed assets are scrapped, the division will incur a loss equal to the net book value of the assets minus the salvage value. This will be equal to the amount that the investment base will be reduced.

Therefore, if fixed assets are included in the investment base at net book value, the division manager will not improve his ROI merely by disposing of fixed assets. Also, if he replaces a fixed asset, his investment base will increase by an amount equal to the net investment (cost of the new asset minus the salvage value of the old). Consequently, there is a reasonable degree of goal congruence between the division and the company with respect to the retirement and replacement of fixed assets if the assets are included in the investment base at net book value.

There is, however, a serious problem with using net book value. This problem results because the investment base of a division will be automatically reduced as the asset ages. Thus the rate of return will increase simply by the passage of time. This situation is aggravated further when a company uses some form of accelerated depreciation (as most large companies do). The amount of depreciation becomes smaller as the assets become older. This can result in very high returns on old assets. Thus, new investments are discouraged because they will reduce a division’s ROI, at least in the short run.

‘Economic value’ concept:

An alternative to using book value might be some kind of “economic value,” but I have not been able to find any satisfactory method for implementing this conceptual alternative. Theoretically, the economic value would be the present value of the future cash flows that will be generated by an asset or a group of assets.

However, from a practical viewpoint, the implementation of this concept has a number of drawbacks. One, of course, is estimating the cash flows and determining a discount rate. Another problem is that we may have circular reasoning if we are not careful. We are trying to measure a division manager’s ROI, which is the profit earned on the investment at his disposal. Yet we are doing this by valuing his investment on the basis of estimated profits from these same investments. Still another problem is that the use of economic value will eliminate the effects of shrewd acquisitions or, conversely, poor acquisition decisions, even though one of the purposes of using ROI is to measure the effect of this type of decision.

The use of economic values is academic, in any case, because no one seems to use any values other than those from the accounting records. I believe that the reason for this is not only our inability to develop meaningful economic values, but also the reluctance of most managers to be evaluated on the basis of values that seem to be artificially created by the economist or accountant. It gives the appearance of “playing games.” Accounting records, on the other hand, are considered “real,” in spite of their obvious limitations.

Transfer pricing:

Intracompany transactions can create a serious problem in the effective operation of ROI financial control. Under the investment-return concept, profit centers deal with each other at “arm’s length” and transfer prices are supposed to approximate the levels that would be paid if the profit centers were independent companies. (This, of course, is necessary when each division is evaluated on the basis of its absolute profitability.) The problem with this approach is that company profits are rarely optimized by the profit centers acting independently. For example:

Division A sells a product to Division B for $100. This price consists of these elements:

To Division B, the entire price is a variable cost and its marketing decisions are made accordingly. To the company, however, the variable costs are $50. The transfer price system, therefore, obscures the relevant costs to the decision maker. (I am assuming that the variable costs will approximate the marginal costs. While this may not always be true, in the short run they will be much closer to the marginal costs than the competitive transfer price.) Even if the decision maker were provided with company variable cost, he would be affecting his profit performance adversely by making decisions on the basis of these costs as long as he was paying a competitive transfer price.

The only time that a transfer price based on a market price represents a company’s marginal cost is the situation arising when the selling division, by selling inside, forgoes the opportunity to sell to the outside. The transfer price, then, represents the opportunity cost of selling to an inside supplier. In practical business operations, this rarely exists except in periods of short supply. As long as a selling division has any unused capacity, the inside sale does not represent a forgone opportunity because a sale could be made to both the inside and the outside customer. Where capacity is limited, the problem of allocating scarce production is even more complex if the company’s profit is to be optimized. Certainly, the use of a transfer price determined by arm’s length bargaining will provide optimum utilization of the scarce goods only by chance.

Capital investment analysis:

At the time that the ROI approach began to be widely adopted, most companies used straight-line depreciation in calculating profits, and the accounting method in evaluating capital investment proposals. As a result, capital investment analysis was consistent with the subsequent measurement of profitability. For example, assume this situation under the accounting method of evaluation in which the proposed investment is $100,000, the estimated annual savings (after tax) is $15,000, and the life of the investment is 10 years:

Using gross book value

Using net book value

When the measurement of profitability is evaluated by gross book value, if the actual results were the same as projected, the ROI would average 5% per year over the life of the investment. If the actual performance each year turned out to be worse than the projected, the division manager’s performance would have been penalized accordingly. Consequently, the ROI method was a means for making a manager responsible for the accuracy of his investment proposals.

On the other hand, using net book value as the basis for evaluation, if the actual results were the same as projected, the investment return would be as shown in Exhibit II. Moreover, the ROI earned over the period would be the same as the return on which the capital expenditure was based. The pattern of this return would be from a low to a high return and, as I explained earlier, this is not really satisfactory for management control because it might discourage investments that would not earn an immediate payback. Over the entire period, however, the average return would be consistent with the capital investment decisions and any deviations from the capital investment estimates would be reflected in the division’s performance.

Exhibit II. Results of Capital Investment Analysis and ROI Using Net Book Value Measurement

Current procedures:

Most progressive companies are using some form of discounted cash flow to make investment decisions and some form of accelerated depreciation to write off assets. As a result of these changes, the ROI earned by a division differs widely from the returns projected in the investment proposals even when actual profits are the same as projected. Let us take the previous example and assume that the company uses the time-adjusted rate of return method for making capital expenditure decisions and the sum-of-the-year’s digits for depreciating fixed assets.

From a table of present values, we find that $15,000 a year for 10 years represents a return of 8% on an investment of $100,000. If we assume that this was equal to the company’s required return, the investment would be approved. The return that will be reflected on the division books, provided the actual results are the same as those projected in the investment proposal, will be far different from the 8%, as shown in Exhibit III.

Exhibit III. Results of Analysis Using Time-adjusted ROI and Sum-of-the-year’s Digits for Depreciation

As you can see in this exhibit, there is little resemblance between the return rate on which the investment decision was based and the ROI earned by the division. This is a serious limitation to the use of ROI for management control. Many division managers will be reluctant to propose capital investments that do not have a rapid early payback, especially those who expect to be moved to another job within a few years. Over the entire period, however, a division manager could miss the projections on which the capital investment decision was based by a large percentage and still earn the ROI projected.

In short, the system could motivate division managers to take incorrect action, yet not measure how effectively they have met the commitments made in their capital investment proposals. Consequently, one of the most important reasons traditionally given for using investment return to measure division performance is no longer applicable in most companies. ROI simply does not provide a means for checking on the accuracy of capital investment proposals.

As we noted earlier, if gross book value is used, the return will average 5% over the life of the investment. It begins, however, with a loss of 3% and, by the tenth year, this will have changed to a profit of 13%. There would still be a reluctance, therefore, to propose investments that did not have an early payback. Also, for reasons mentioned previously, the use of gross book value so completely distorts the return earned on a replacement investment that I do not see how this method can be used to measure the effectiveness of a division manager in making capital expenditures.

Correcting the deficiencies

Some of the preceding technical limitations just described can be avoided by modifying the ROI system. My purpose here is to describe these modifications and assess their usefulness in correcting the deficiencies of ROI.

‘Residual income’ method:

Perhaps the most important adaptation of the ROI concept is the “residual income” approach. In this method, a division is measured by its actual profits minus a prescribed charge for the actual investment or capital it employs. For example, assume that a division had a profit potential of $1,500,000 per year and a budgeted investment of $10,000,000. The company has a cutoff rate of 10%. The profit objective of the division is expressed as 10% on investment plus $500,000—i.e., $1,500,000 minus .10 ($10,000,000). If the actual investment for a year were $11,000,000 and the division earned $1,450,000, the actual residual income would be $350,000—i.e., $1,450,000 minus .10 ($11,000,000)—or $150,000 less than objective.

There are three advantages to the residual income method:

1. It makes it practicable to use different rates of return for different types of assets. For example, the marginal borrowing rate can be used for inventories while the long-term cost of capital can be used for investments in fixed assets. Further, different rates can be used for different types of fixed assets to take into account different degrees of risk.

2. It can be used to require the same type of asset to earn the same return, regardless of the profitability of the particular division. Thus it establishes consistency among divisions with respect to the desirability of investing in new assets.

3. It avoids the problems that occur when a division has a very high investment return. Such a division can lower its ROI if it makes almost any new investment. It can, however, increase its residual income by any investment that yields a profit higher than the required percentage.

Although the residual income method is much better than the investment-return approach, it still does not solve one of the basic problems with ROI control. Fixed assets must still be included in the investment base at some value, and the problem of which value to use still exists. If gross book value is used, residual income can be increased by scrapping or replacing assets, regardless of the economic impact to the company. If net book value is used, residual income will increase automatically through the passage of time. Furthermore, new investments, with a satisfactory average rate of profit, will often reduce the residual income of a division in the early years. Residual income, therefore, is no better than ROI as a method of controlling investments in fixed assets.

In any case, it does not appear that the residual income method is being widely used. The Mauriel and Anthony studies showed that only 7% used the residual income method exclusively; 20% of the respondents used residual income in conjunction with ROI; and 60% used ROI exclusively.

‘Annuity’ depreciation:

The problem of inconsistency between the return indicated in capital investment analyses and that shown on the division profit statements can be partially overcome by using the “annuity” method of depreciation. This is the opposite of accelerated depreciation in that the annual amount of depreciation increases over time. Annuity depreciation operates on the principle that the rate of return is a constant percentage of the net book value, and that depreciation represents the return of capital.

Consequently, in the early years, the proportion of earnings required to maintain the average return will be relatively high and the return of principal relatively low. As the book value decreases, the profits required to maintain the rate of return will decrease and the return of principal (the depreciation) will increase. (In fact, it operates exactly like paying off a house mortgage.) Exhibit IV is an example of annuity depreciation.

Exhibit IV. Hypothetical Example of “Annuity” Method of Depreciation

Where the cash flows are even, the annuity method of depreciation is entirely consistent with the discounted cash flow method of evaluating capital investment proposals. In actual business situations, the cash flows are practically never even. However, in those cases where the cash flows are not even, the annuity method of depreciation will approximate the discounted cash flows much better than accelerated or straight-line depreciation.4

The most serious obstacle to using annuity depreciation is management’s reluctance to utilize a method of measuring division performance which is different from that used in the accounting records. As I indicated earlier, most companies are extremely reluctant to deviate from accounting records in measuring ROI. To my knowledge, no company uses the annuity method of depreciation in measuring division performance.

Two-step pricing:

The intracompany pricing problem can be corrected, at least partially, by using a two-part price for transfers between divisions. This price consists of (1) an amount per unit equal to the standard variable cost; and (2) a total monthly amount equal to the fixed costs plus a return on the investment dedicated to the production of the product being transferred.

For example, assume these conditions: Division A reserves capacity to produce 1,200,000 units of Product X per year for Division B; the variable cost per unit is $3.50; the fixed costs per month are $100,000; the investment required to produce 1,200,000 units is $10,000,000; and the return that should be earned on this investment is 6%. (If competitive prices are available, the ROI will be the same as that earned at competitive prices. If no competitive prices are available, it will usually be the objective return for the selling division.) The price would be $3.50 for each unit sold by A to B plus a monthly charge of:

Thus the variable cost to Division B will be the same as that of the company and, consequently, the company profit will be optimized by optimizing the division profit.

There are several problems with implementing this method of intracompany pricing. In the first place, how do you determine the capacity to be dedicated to produce a particular part, and what happens if a division requires more capacity than it initially requested? These problems can be particularly difficult if both an inside division and an outside customer are involved when there is a shortage of capacity. Another problem occurs when a part is transferred between two or more divisions of a company. Keeping track of fixed and variable expenses can sometimes become difficult. Moreover, standard variable costs will not always be the marginal costs of the company (although they will be a lot closer to marginal costs than a negotiated transfer price). These problems, together with a reluctance on the part of managers to deviate from the “arm’s length bargaining” price, have resulted in only a very limited use of this method of transfer pricing.

If a company were to use (a) the residual income method of division profit objectives, (b) annuity depreciation, (c) fixed assets at net book value, and (d) composite depreciation, it would be possible to combine the measurement of profit and investment without motivating the manager to take uneconomic action. To do this, however, would take drastic changes in the division’s accounting procedures, which most companies seem reluctant to make. Furthermore, I question what would be gained over controlling investment and profits separately. The system would not provide a check of how well the manager was investing company funds; profits for performance measurement would differ significantly from the accounted profits because few companies would use annuity depreciation for financial statement purposes; and there would be no basis for evaluating profitability that is comparable among divisions or with outside companies.

However, since no company adapts ROI in the ways just described, all investment return systems in use today have many more limitations than these. At best, therefore, ROI systems do not accomplish what they were designed to do and, in every instance, encourage division managers to take some actions contrary to the overall interests of the company.

Implementation constraints

In addition to the technical problems described in the preceding part of this article, the ROI system is also subject to some rather critical limitations in application. The idea of establishing a profit objective for each division and of measuring actual performance against this objective each month, quarter, or year is an enticing concept. However, there are three problems with implementing this concept in most companies.

First, it can be very difficult to set equitable annual profit objectives. The economic environment of the typical profit center is so complex that it is often impossible to answer, with any degree of reliability, the question: “What rate of return (or what profit) should this division earn this year?” Yet the effectiveness of the entire system depends on a reliable answer to this question. If you cannot set an equitable goal, how can you evaluate performance against this goal?

Second, the annual accounted profit can be a poor measure of what has been accomplished during any relatively short period of time. For example, a new product could have been developed that will affect future performance significantly; or, organization changes could have occurred that will affect future performance. Neither of these changes will be reflected in current profitability.

Third, it is often difficult to assign responsibility for a deviation from the profit objective. So many of the causes of a deviation from objective are either unknown or imperfectly known that many times it is impossible to classify variances from objectives into those which are controllable by the division manager and those which are not. Thus, even if you started with a fair objective at the beginning of the year, it might not be possible to evaluate performance on the basis of this objective. If conditions had changed from those assumed when the profit objective was approved, it might not be possible to assign responsibility for those changes.

The technical limitations of the ROI system will apply to every decentralized company and can more or less be “proved” to exist. That is to say, while a company president may state that “my people would not act to increase division ROI at the expense of company profit,” there can be no question that certain actions will increase division return at the expense of companywide profits. The implementation limitations of ROI, however, may not apply to every decentralized company. Furthermore, the impact of these limitations will be different not only in different companies, but also within divisions of the same company.

Time-span evaluations:

Perhaps the major problem with setting profit objectives and evaluating performance against those objectives is that one year is often too short a period to evaluate a task as complex as managing a profit center. Although we are just beginning to use the concept of time span in our management control theory, it is easy to show that certain types of jobs cannot be evaluated within a one-year period.

For example, it may take several years to evaluate a manager responsible for a new product innovation. The ROI of annual profit objectives and quarterly evaluations of performance against those objectives will not be very useful in evaluating divisions with time spans of evaluation longer than one year. The longer the time span of a division manager, the more difficult it will be to evaluate his performance using ROI.

Familiarity with operations:

Clearly related to time span is top management’s familiarity with division operations. The more familiar top managers are with the operating problems of a division, the better they will be able to evaluate performance. This statement, of course, is self-evident. What is frequently missed, however, is the importance of this familiarity. Because of the technical limitations of the ROI system, its use can result in serious mistakes unless the top managers are intimately familiar with division operations. This familiarity is really the only adequate defense against the dysfunctional aspects of ROI.

In fact, familiarity with operating problems is one way to reduce the time span required to evaluate division management. For example, if the quality of a division manager’s decisions can be evaluated directly, it would not be necessary to wait until the effect of his decisions were reflected in profits.

In my experience, many business executives seem to believe that decentralized profit center control systems, in themselves, will provide adequate control. According to their view, all management needs to do is to set objectives and, thereafter, manage by exception. If a division is making its objective, no action is necessary. Management time and effort is spent on those divisions that are “in trouble” (i.e., considered generally to be those divisions failing to meet their profit objectives).

The management belief that control systems per se will provide adequate control is an interesting theory. For the reasons stated earlier, however, this just does not work out in many companies. Monthly or quarterly accounting figures, in themselves, simply will not provide the information required for month-to-month control. Consequently, management must either become familiar with division problems or face the fact that it must delegate responsibility to the division manager over a period of time that may extend into several years.

Profit objectives:

Management control theory mistakenly considers the profit objective to be within a narrow band. If the profit objective is met, no management action is required. Conversely, where it is not met, management takes the appropriate action to correct the unsatisfactory situation. The concept of a narrow band of satisfactory performance probably comes from standard cost systems. In manufacturing activities, the standard cost usually represents both the marginal acceptable performance and the average expected performance. This narrow range of acceptable performance is possible because we are able to measure, to quite close tolerances, how much labor and material are required to produce a given part. Thus an operating variance means that management action may be required.

However, contrary to management control theory, in actual practice, a profit objective usually represents a “high level of performance,” so there can be a wide band between profit objective and submarginal performance. Management action is not required, therefore, every time a division misses its profit objective. Thus the profit objective, even at best, will not be an exact instrument for financial control. In those instances where objectives cannot be set accurately, where profit measurements are inexact, and where the actual conditions vary significantly from those assumed when the objective was set, the profit objective can be a very unsatisfactory basis for measuring division performance.

Developing New Systems

The principal purpose of this article has been to point out the deficiencies in ROI financial control. I believe that it is important to recognize these deficiencies so that considerable research effort can be devoted to developing satisfactory new financial control systems. Consequently, I would not be presumptuous enough to imply that I know how to correct the deficiencies we have noted. Nevertheless, I believe that some actions can be taken immediately to improve management control systems.

First let me state that I believe that an alternative to the ROI system is not the recentralization of profit responsibility. In modern business, it is necessary to delegate profit responsibility down the line because only the manager with profit responsibility will have the incentive (or perhaps the necessary information) to make correct economic trade-offs. Most large businesses are much too complex to make these trade-offs at the top (or near top) levels of the organization. This being the case, it is necessary to have a system of management control where profit responsibility has been decentralized.

Methods available now

One of the responses consistently given to criticism of the ROI system is: “I agree it is not perfect but it is the best system available.” This is the crux of the problem. Are better methods already available, or do we have to continue using the ROI system until new research provides better answers than are now available? I believe there are three methods currently available that should be adopted pending further improvements from new research.

First, I believe we should face the fact that we cannot control the investment in fixed assets through either the investment return or the residual income method.

Consequently, we should stop evaluating division performance on the basis of a ratio (profit divided by investment) or by charging management with a cost for the fixed assets being used. We should control profits and fixed assets separately. The division manager’s profit performance should be based on the amount of profits he has earned compared to some type of objective. Fixed assets should be controlled by capital investment analysis procedures and post-completion audits. It seems to me that a company has everything to gain and nothing to lose in controlling these two separately. ROI does not provide any control and only obscures the fact that there is no real control. At the same time, the attempt to control profit and investment by a single ratio results in all of the problems that I described earlier.

I think we should also exclude from division performance depreciation and other changes representing allocations of “sunk” costs over time. Since the money has already been spent, and allocations are more or less arbitrary, I do not see what is to be gained by including them in a performance measurement system. Until better systems are developed, therefore, I recommend that division performance be measured on the basis of revenue minus out-of-pocket costs. In addition to out-of-pocket costs, a charge should be made for controllable current assets (e.g., inventories) equal to the company’s marginal cost of money tied up in these assets to provide the necessary incentive to maintain proper control over them.

Second, the intracompany transfer pricing system should provide the divisions that finally sell to the outside customer with the information and incentive to maximize company profits.

This will mean the use of some form of the two-step transfer price described earlier. This may appear to be an artificial pricing mechanism. The use of full costs plus profits, however, can result in serious suboptimization of company profits. The fact remains that two divisions of the same company are not independent. Thus to treat them as though they were independent means that a company is actually giving up some of the benefits of being a unified organization.

Third, some thought should be given to the time span required to evaluate division performance.

Where evaluation requires a period longer than a year, it seems to me that management should face up to this situation and extend this period. The concept of time span is relatively new, but some research is currently being conducted in this field. Hopefully, we can expect some results in the next few years. In the meantime, it seems to me that we should not necessarily be tied down to an absolute accounting period in our formal control systems. Again, management has nothing to lose and everything to gain. If a division manager cannot be evaluated in a year, it only fragments the evaluation to use a system based on an annual time span.

I believe that it is most important to get away from blindly using the ROI system and to develop procedures that best satisfy the requirements of each individual company. In the management control field, I think we should recognize that ROI financial control was a very useful innovation when it was first developed, but that it is now obsolete.

1. Executive Control—The Catalyst (New York, John Wiley & Sons, Inc., 1961), p. 195.

2. Jerome, op. cit., p. 200.

3. See “Misevaluation of Investment Center Performance,” HBR March–April 1966, p. 98.

4. For a more detailed description of this concept, see Robert N. Anthony, “Accounting for Capital Costs,” in Robert N. Anthony, John Dearden, and Richard F. Vancil, Management Control Systems: Cases and Readings (Homewood, Illinois, Richard D. Irwin, Inc., 1965).

A version of this article appeared in the May 1969 issue of Harvard Business Review.

In what way can the use of ROI as a performance measure for investment centers?

In what way can the use of ROI as a performance measures for investment Centres lead bad decisions How does the residual income approach overcome this problem?

What is residual income and how it is applied in evaluating new investment project when compared to ROI?

Why does the balanced scorecard include financial performance measures as well as measures of how well internal processes are doing?