2. Economic Value Added and its characteristics

This chapter presents the main theory about EVA and shows some empirical findings around the concept in financial literature. The last section 2.3 tries to present what the theory of EVA means in practice for companies.

2.1 The main theory behind EVA

EVA measures whether the operating profit is enough compared to the total costs of capital employed. Stewart defined EVA (1990, p.137) as Net operating profit after taxes (NOPAT) subtracted with a capital charge:

EVA = NOPAT - CAPITAL COST Û

EVA = NOPAT - COST OF CAPITAL x CAPITAL employed (1)

Or equivalently, if rate or return is defined as NOPAT/CAPITAL, this turns into a perhaps more revealing formula:

EVA = (RATE OF RETURN - COST OF CAPITAL) x CAPITAL (2)

Where:

Rate of return = Nopat/Capital

Capital = Total balance sheet minus non-interest bearing debt in the beginning of the year

Cost of capital = Cost of Equity x Proportion of equity from capital + Cost of debt x Proportion of debt from capital x (1-tax rate). Cost of capital or Weighted average cost of capital (WACC) is the average cost of both equity capital and interest bearing debt. Cost of equity capital is the opportunity return from an investment with same risk as the company has. Cost of equity is usually defined with Capital asset pricing model (CAPM). The estimation of cost of debt is naturally more straightforward, since its cost is explicit. Cost of debt includes also the tax shield due to tax allowance on interest expenses. This derivation of equity cost and WACC is explained later in detail with chapter 4.2 (Company B's EVA).

If ROI is defined as above (after taxes) then EVA can be presented with familiar terms to be:

EVA = (ROI - WACC) x CAPITAL EMPLOYED (3)

The idea behind EVA is that shareholders must earn a return that compensates the risk taken. In other words equity capital has to earn at least same return as similarly risky investments at equity markets. If that is not the case, then there is no real profit made and actually the company operates at a loss from the viewpoint of shareholders. On the other hand if EVA is zero, this should be treated as a sufficient achievement because the shareholders have earned a return that compensates the risk. This approach - using average risk-adjusted market return as a minimum requirement - is justified since that average return is easily obtained from diversified long-term investments on stock markets. Average long-term stock market return reflects the average return that the public companies generate from their operations.

EVA is based on the common accounting based items like interest bearing debt, equity capital and net operating profit. It differs from the traditional measures mainly by including the cost of equity. Mathematically EVA gives exactly the same results in valuations as Discounted cash flow (DCF) or Net present value (NPV) (Stewart 1990, p.3 and Käppi 1996), which are long since widely acknowledged as theoretically best analysis tools from the Shareholders perspective (Brealey & Mayers 1991 p.73-75). These both measures include the opportunity cost of equity, they take into account the time value of money and they do not suffer from any kind of accounting distortions. However, NPV and DCF do not suit in performance evaluation because they are based exclusively on cash flows. EVA in turn suits particularly well in performance measuring. Yet, it should be emphasized that the equivalence with EVA and NPV/DCF holds only in special circumstances (in valuations) and thus this equivalence does not have anything to do with performance measurement. This peculiar characteristic of EVA is explained later in detail.

2.1.1 The background of EVA

EVA is not a new discovery. An accounting performance measure called residual income is defined to be operating profit subtracted with capital charge. EVA is thus one variation of residual income with adjustments to how one calculates income and capital. According to Wallace (1997, p.1) one of the earliest to mention the residual income concept was Alfred Marshall in 1890. Marshall defined economic profit as total net gains less the interest on invested capital at the current rate2. According to Dodd & Chen (1996, p.27) the idea of residual income appeared first in accounting theory literature early in this century by e.g. Church in 1917 and by Scovell in 1924 and appeared in management accounting literature in the 1960s. Also Finnish academics and financial press discussed the concept as early as in the 1970s. It was defined as a good way to complement ROI-control (Virtanen 1975, p.111). Knowing this background many academics have been wondering about the big publicity and praise that has surrounded EVA in the recent years. The EVA-concept is often called Economic Profit (EP) in order to avoid problems caused by the trademarking. On the other hand the name "EVA" is so popular and well known that often all residual income concepts are often called EVA although they do not include even the main elements defined by Stern Stewart & Co. For example, hardly any of those Finnish companies that have adopted EVA calculate rate of return based on the beginning capital as Stewart has defined it, because average capital is in practice a better estimate of the capital employed. So they do not actually use EVA but other residual income measure. This insignificance detail is ignored later on in order to avoid more serious misconceptions. It is justified to say that the EVA concept Finnish companies are using corresponds virtually the EVA defined by Stern Stewart & Co.

In the 1970s or earlier residual income did not got wide publicity and it did not end up to be the prime performance measure in great deal of companies. However EVA, practically the same concept with a different name, has done it in the recent years. Furthermore the spreading of EVA and other residual income measures does not look to be on a weakening trend. On the contrary the number of companies adopting EVA is increasing rapidly (Nuelle 1996, p.39, Wallace 1997, p.24 and Economist 1997/2). We can only guess why residual income did never gain a popularity of this scale. One of the possible reasons is that Economic value added (EVA) was marketed with a concept of Market value added (MVA) and it did offer a theoretically sound link to market valuations. In the times when investors demand focus on Shareholder value issues this was a good bite. Perhaps also pertinent marketing by Stern Stewart & Co. had and has its contribution.

2.1.2 Market Value Added defined

EVA is aimed to be a measure that tells what have happened to the wealth of shareholders. According to this theory, earning a return greater than the cost of capital increases value (of a company), and earning less decreased value. For listed companies Stewart defined another measure that assesses if the company has created shareholder value. If the total market value of a company is more than the amount of capital invested in it, the company has managed to create shareholder value. If the case is opposite, the market value is less than capital invested, the company has destroyed shareholder value. Stewart (1990,153) calls that difference between the company's market and book value as Market Value Added or MVA™ for short3.

MarkeT Value Added =

company's total Market Value - capital invested

and with simplifying assumption that market and book value of debt are equal, this is the same as:

MarkeT Value Added =

Market Value of Equity - Book Value of Equity (3)

Book value of equity refers to all equity equivalent items like reserves, retained earnings and provisions. In other words, in this context, all the items that are not debt (interest bearing or non-interest bearing) are classified as equity.

Market value added is identical by meaning with the market-to-book -ratio. The difference is only that MVA is an absolute measure and market-to-book -ratio is a relative measure. If MVA is positive that means that market-to-book -ratio is more than one. Negative MVA means market-to-book -ratio less than one.

According to Stewart Market value added tells us how much value company has added to, or subtracted from, its shareholders' investment. Successful companies add their MVA and thus increase the value of capital invested in the company. Unsuccessful companies decrease the value of the capital originally invested in the company. Whether a company succeeds in creating MVA (increasing shareholder value) or not, depends on its rate of return. If a company's rate of return exceeds its cost of capital, the company will sell on the stock markets with premium compared to the original capital (has positive MVA). On the other hand, companies that have rate of return smaller than their cost of capital sell with discount compared to the original capital invested in company. Whether a company has positive or negative MVA depends on the level of rate of return compared to the cost of capital. All this applies also to EVA. Thus positive EVA means also positive MVA and vice versa. Stewart (p. 153) defined in his book the connection between EVA and MVA.

MarkeT Value Added = Present value of all future eva (4)

Market value added is equal to present value of all future EVA. Increasing EVA a company increases its market value added, or in other words increases the difference between company's value and the amount of capital invested in it.

The relationship with EVA and MVA has its implications on valuation. By arranging the formulas above we find a new definition of the value of company:

Market Value of Equity =

Book Value of Equity + Present value of all future eva (5)

Following figure will illustrate this relationship between EVA and MVA:

Figure 1 Company's market value depends directly on its future EVA.

The phenomenon with rate of return and MVA is in one sense similar to the relationship between the yield and market value of a bond. If the yield of a bond exceeds the current market interest rate (cost of capital) then the bond will sell at a premium (there is positive EVA and so the bond will sell at positive MVA). If the yield of a bond is lower than the current market interest rate then the bond will sell at discount (there is negative EVA and so the bond will sell at negative MVA).

If the net assets or "capital" in the EVA formula (formula 2) reflected the current value of a company's assets and if the "rate of return" reflected the true return, then there would not be much questioning about the theory between EVA and MVA. After all, nobody questions the above connection between the market value, face value, interest rate and yield of a bond (obviously not since it hold almost perfectly also in practise). But with MVA and EVA things are little bit more complicated. The term "capital" in formula 2 does not reflect the current value of assets, because the capital is based on historical values. Nor does the "rate of return" reflect the true return of the company. All accounting based rate of returns (ROI, RONA, ROCE, ROIC) fail to assess the true or economic return of a firm, because they are based on the historical asset values, which in turn are distorted by inflation and other factors (Villiers 1997, p.287). Stewart defines his rate of return as return on beginning capital and as return after taxes but these adjustments do not affect the problems attached to accounting rate of return. The shortcomings of accounting rate of returns and the current research on the subject are presented in detail in next section (2.2.1.).

The valuation formula of EVA (formula 5) however is always equivalent to Discounted cash flow and Net present value, if EVA is calculated as Stewart presents. Thus the above valuation formula (formula 5) gives always the right estimate of value (same as DCF and NPV) no matter what the original book value of equity is. This holds true even though capital is not an unbiased estimate of current value of assets and rate of return is not an unbiased estimate of the true return. That is because an increase in book value (formula 5) decreases the periodic EVA-figures (and of course a decrease in book value increases EVA-fig.) and these changes cancel each other out. Also this phenomena will be discussed more in next section (2.2.1).

2.2 A review of EVA as performance measure and as a yardstick of wealth creation

2.2.1 The discrepancy in accounting rate of return (ROI) and EVA

Every project that a firm undertakes should have positive Net present value (NPV) in order to be acceptable from the shareholders point of view. This means that a project should have internal rate of return4 bigger than the cost of capital. With practical performance measuring the internal rate of return can not be measured and some accounting rate of return is used instead to estimate the rate of return to capital. Typically this rate of return is some form of return on investment (ROI). Unfortunately any accounting rate of return can not on average produce an accurate estimate of the underlying true rate of return. Following example illustrates this problem, which is more thoroughly and with stronger theoretic background discussed below. The example presents an investment project with initial investment of 1200, duration of 8 years, constant gross profit of 210, IRR of 11% and with no salvage value.

Table A: Example how ROI estimates (both in different years and on average) the return of an investment producing a IRR of 11%.

	Year	1	2	3	4	5	6	7	8
Cash flows
Investment		-1200
Gross margin		210	210	210	210	210	210	210	210
Total Cash flow		-990	210	210	210	210	210	210	210
Depreciation		-150	-150	-150	-150	-150	-150	-150	-150
Operating income		60	60	60	60	60	60	60	60
Balance sheet
Beginning assets		1200	1050	900	750	600	450	300	150
Ending assets		1050	900	750	600	450	300	150	0
Accounting returns
ROI (beginning)		5,00 %	5,71 %	6,67 %	8,00 %	10,00 %	13,33 %	20,00 %	40,00 %
ROI (average)		5,33 %	6,15 %	7,27 %	8,89 %	11,43 %	16,00 %	26,67 %	80,00 %
True return and Net present value
IRR		11,0 %	11,0 %	11,0 %	11,0 %	11,0 %	11,0 %	11,0 %	11,0 %
NPV (WACC 10%)		29
Different averages of ROIs			Normal average	Harmonic mean	Geo-metric mean	Normal mean weighted with assets
ROI based on beginning capital			13,59 %	8,9 %	10,6 %	8,9 %
ROI based on average capital			20,22 %	10,0 %	13,0 %	10,0 %

Figure 2: How ROI estimates the return of an 8-year-project in different years. The true return or IRR of the project is 11% (shown as a vertical line).

As the above example indicates, ROI is a poor indicator of the true rate of return of the project. The Table A and the Figure 2 illustrate how ROI underestimates the IRR in the beginning and overestimates it in the end on the period. In the remaining study this phenomenon is called wrong periodizing. Besides that ROI periodizes the rate of return wrongly in this example and it also on average fails to estimate the true rate of return of the project. That can be seen from the different averages of ROI in the bottom of the Table A. None of them is the same as IRR. In this case ROI underestimates5 the true return. In the real life inflation increases the cash flows compared to the initial investment and thus ROI might as well overestimate the true return.

The wrong periodizing is with a real project perhaps even fiercer than in the above example. That is because usually in the real life projects the positive cash flows are generated only some time after the beginning of the period. For example investment in a new plant or machinery starts to generate positive cash flows only after construction and installation phase. It also takes some time to reach the full potential of new machines and it might take some time to establish new product in the markets.

However a company is typically a continuous stream of investments and not a single big investment. Therefore the problem of wrong periodizing of accounting rate of return is with performance measurement not as big a problem as with a single investment. Furthermore, a company has also a big proportion of current assets that reduce the problem of wrong periodizing. That is because there are approximately as much current assets in the beginning as in the end of the investment period. However the wrong periodizing is a problem. Companies can have a big proportion either very old or young assets. It is seldom the case that there are equal proportions of old, young and middle aged assets in a company's balance sheet. Thus if a company has a lot of new assets, new investments, it its likely to have low ROI although its true rate of return were sufficient. In the opposite case, a company has very little new investments compared to the major investments made in the past. This kind of example can be e.g. a very old paper mill: Since the original investment is depreciated, the assets are very small. Therefore a moderate operating cash flow might produce a very high ROI although the true return for the whole investment period is even lower than the cost of capital. This kind of situation might give the management wrong signals of the true profitability of a business. Thus it might lead to either overinvestments in mature businesses or underinvestments in profitable businesses. Furthermore, on the basis of the above example (Table A and Figure 2), it is easy to see that ceasing investments leads to increase in ROI in the short run.

In addition to wrong periodizing ROI is also otherwise a poor measure of company's true rate of return. The discrepancy between the accounting rate of return and the true return is well documented in economic literature. Harcourt (1965), Salomon and Laya (1967), Livingston and Salomon (1970), Fischer and McGowan (1983) and Fisher (1984) concluded that the difference between accounting rate of return and the true rate of return is so large that the former can not be used as an indication of the latter (REF De Villiers 1997, p.286-287). The effect of inflation on the discrepancy was addressed by Salomon and Laya (1967), Kay (1976), Van Breda (1981) Kay and Mayer (1986) and De Villiers (1989). They have shown that inflation exacerbates the discrepancy between accounting and true return. (REF De Villiers 1997, p.286-287) Although inflation strengthens the discrepancy, it should be pointed out that accounting rate of return is not, on average, equal to the true rate of return even with no inflation.

Salomon and Laya (1967) studied the accounting rate of return (ARR) and the extent to which it approximates the true return measured with IRR. The IRR of a project can be measured, but because the projects constituting a firm are usually not visible, the true yield of a firm is unknown (Salomon and Laya, 1967, p. 157). The authors therefore studied a theoretical firm made up from projects with a known IRR, and found that the ARR of the firm differs from the IRR of the projects underlying the theoretical firm. The authors also show by means of a numerical simulation that inflation increases the ARR of a firm when IRR is being held at constant. (REF De Villiers 1989, p. 494-495)

De Villiers (1989) studies the relationship between accounting and true rate of return with different asset structures. Typically firms can have three different type of assets: Current assets (inventories and receivables), Depreciable assets (e.g. machinery&equipment and buildings) and Non-depreciable assets (e.g. land and stocks). De Villiers (1989) finds that if a firm had nothing but current assets, ROI (on average) would equal IRR. However, the more a firm has depreciable assets (ceteris paribus), the more ROI overstates IRR. On the other hand the more firm has non-depreciable assets (ceteris paribus) the more ROI understates IRR. In the real world companies have assets of all these three kinds and their relative proportions determine whether ROI underestimates or overestimates IRR (and true rate of return). De Villiers (1989) also presents that even if the assets are valued at their current value (and not at their historical value) there is still some discrepancy between ROI and IRR. In other words when the understatement of asset value (caused by inflation and historical values) is eliminated there is still discrepancy between ROI and IRR that can thereby be ascribed to a deficiency in the accounting profit only. (De Villiers 1989, p.502-503) De Villiers concludes that accounting rates of return of firms with different asset structures are not comparable.

Alongside with inflation rate and asset structure, also the length of investment period affects the discrepancy between ROI and IRR. Other factors being constant, the longer investment period (economic life of assets) the bigger is the discrepancy between ROI and IRR. This is obvious since long investment period gives inflation time to distort asset values. The effect of the project duration to the discrepancy is shown in the article of De Villiers (1997, p.293-294).

Since EVA is calculated from the accounting based numbers and some version of accounting return is used in calculating EVA, it is obvious that all the discrepancies mentioned above affect also EVA. If ROI overstates IRR then EVA also overstates the real shareholder value added. De Villiers (1997) demonstrates with numerical examples how big these distortions can be. He also suggests the use of a modified concept of EVA called adjusted EVA (or AEVA) in order to radically decrease these discrepancies. The adjusted EVA is simple using current value of all assets in calculating the accounting rate of return (ROI). De Villiers pointed out that one should not use market values of equity in calculating EVA as so often is done. Using market value of equity would be circular reasoning and lead to EVA of zero. Instead current value (market value) of individual assets produce much more sound result, but they are admittedly often either very difficult or even impossible to estimate. The use of current value of assets does not however eliminate the discrepancy wholly but it does diminish it to a fraction of original discrepancy.

Storrie & Sinclair (1997) present also that EVA based on historical values can be somewhat misleading. They first demonstrate that the valuation formula of EVA is theoretically exactly the same as the valuation formula of discounted cash flow (DCF) (Proved also by Käppi 1996). After that Storrie & Sinclair also prove mathematically that this equivalence is due to the fact that the book value in EVA valuation formula is irrelevant in determining value. That is because an increase in "book value of equity" (formula 5 below) decreases the periodic EVA-figures ("present value of future EVA") and these changes cancel each other out.

Market value of equity = Book value of equity + present value of future EVA (5)

Book value of equity affects the periodic EVA figures in future via capital costs: If book value of equity is too high then the capital costs in future are also too high and the periodic EVA values too low. These opposite changes in the two terms cancel each other and thus the market value of equity is always the same no matter of the original book value. This is quite simple to demonstrate with an example:

Suppose that a company does an asset revaluation of +100 and thus increases its book value of equity from 500 to 600. The increase in net worth is naturally only an accounting trick and does not affect the market value of company. Let us examine the impact of this trick to above EVA-valuation formula (formula 5). The additional book value of 100 increases periodical capital costs with 100 x WACC (let us assume that this additional book value of 100 is undepreciable, which makes the example easier). If WACC is assumed to be 10%, then the increase in periodic capital costs is 10. How much does this periodic increase in capital costs decrease the present value of EVA? Well, if the additional capital cost decreases periodic EVA by 10 with each year then the whole impact can be calculated as a present value of this 10. The present value of this 10 is 10/0,1 = 100 (The Gordon model: the present value of infinite and constant cash flow: PV= D/r). Hence the decrease in present value of EVA (-100) is with absolute value exactly the same as the increase in book value of equity (+100). Therefore this action does not affect the market value of equity calculated with EVA.

As we can see the decrease in the present value is exactly as big as the increase in book value, so the initial book value does not matter in valuation. More generally proofed:

Change in present value of future EVA = (Change in book value x capital cost)/capital cost6 = Change in book value (6)

The situation does not change even if the change in book value was depreciable. Then the additional depreciation and additional capital costs correspond together the change in book value.

This is the reason why a measure like EVA based on accounting items can produce theoretically equivalent result with discounted cash flow although we know that accounting based measures and accounting based rate of returns are somewhat distorted. According to Storrie & Sinclair (1997):

The mathematical equivalence is achieved because the EVA formula is a modified version of a standard DCF formula within a mathematical construct in which all of the adjustments in the EVA formula to the DCF must result net to zero. The result of this construct is that it does not matter what beginning capital base is used in an EVA valuation - the result value will always be identical.

EVA valuation formula gives the true value of a firm no matter how the accounting is done. This is achieved with combining income statement and balance sheet. Double entry bookkeeping ensures that everything must add up and that accounting numbers have some connection with economically meaningful variables such as cash flow and dividends. This discipline applies however only when profit is computed on a "comprehensive income" basis:

Opening book value of equity

+ Accounting profit

- Dividends (less new issues of equity)

= Closing book value of equity

For this relationship to hold, profit must include all valuation adjustments affecting the balance sheet. In some countries it is possible to violate against this principle. (O'Hanlon & Peasnell, 1996)

Although in valuation the capital base does not matter, it might cause harm in performance measurement because the periodic values of EVA are distorted. This distortion can be abolished almost entirely by using current value of assets in calculating capital costs, but again this might be time-consuming and difficult and it might not pass a prudent cost benefit analysis in practical business situation (Dodd&Chen 1996, p.28). It should also be noted that in practice EVA seldom corresponds DCF, because any adjustment made to EVA abolishes the mathematical equivalence (Storrie&Sinclair 1997, p.5).

Also the original EVA consulting company Stern Stewart & Co has noticed and reacted to the distortions in periodic EVA figures. The company recommends that after introducing a simple definition of EVA, the concept can be refined to the degree that makes sense taking into account both the costs and benefits of complicating the model. According to Stern Stewart, two most important ways to decrease accounting distortions are introducing a modified depreciation schedule or imposing a level capital charge throughout the life of the asset. Either of these prevents EVA from increasing simply because an asset is growing older. (Kroll 1997, p.105) The level capital charge means probably that the sum of depreciation plus capital cost of an asset is the same every year during the economic life of the asset in question. Normally the depreciation is the same every year (straight-line depreciation) and thus the sum of depreciation and capital costs is big in early years and diminishes towards the end.

2.2.2 Some evidence on the correlation between EVA and share prices

As already presented (in chapter 2.1.2.) according to the EVA-theory the market value of a company is its book value plus the current value of future EVA (formula 5). This strict relationship between EVA and the market value of a company suggests that EVA drives the market values of shares. This relationship between EVA and MVA has been studied in the recent years in many studies with many methods - and with different results.

Stewart (1990, p.215 - 218) has first studied this relationship with market data of 618 U.S. companies. Stewart presents the results in his book "The quest for value". He states that EVA and MVA correspond each other in reality quite well among US companies (the data was from late 1980's). Only the relationship between negative EVA and negative MVA does not hold very well. According to Stewart, this is because the potential of liquidation, recovery, recapitalization, or takeover sets a floor on a company's market value (Stewart p.217). For example with companies which have a lot of fixed assets this is quite easy to understand. Market value will always reflect the value of assets even though the company has very low or negative rate of return (and so theoretically it should sell a lot below book value). That is because the company can always be liquidated; the owners have an option to liquidate the assets if the return looks week also in the future. On the other hand markets do not believe that the weak returns can go on forever. Markets are expecting a chance, an improvement, in the long run. If EVA is positive, the relationship is more direct. Then the market valuation happens on the basis of return and growth potential and not on the basis of liquidation or recovery value. Stewart finds also that MVA and EVA correspond each other best when we talk about changes in EVA and MVA and not the absolute levels. Changes in EVA and MVA are not affected so much by accounting distortions and inflation than the absolute values.

Lehn and Makhija (1996) study EVA and MVA as performance measures and signals for strategic change. Their data consists of 241 U.S. companies and cover years 1987, 1988, 1992 and 1993. The researchers first find out that both measures correlate positively with stock returns and that the correlation is slightly better than with traditional performance measures like return on assets (ROA), return on equity (ROE) and return on sales (ROS). Additionally they study how companies' performance, as measured in terms of EVA and MVA, affect on the CEO firings. Finally they examine the relationship between EVA/MVA and corporate focus. Lehn and Makhija find an inverse relation between EVA/MVA and abnormal CEO turnover. They also find that firms with greater focus on their business activities have significantly higher MVA than their less focused counterparts. Lehn and Makhija conclude that their results suggest EVA and MVA to be effective performance measures that contain information about the quality of strategic decisions and serve as signals of strategic change.

Uyemura, Kantor and Pettit (1996) from Stern Stewart & Co present findings on the relationship between EVA and MVA with 100 bank holding companies. They calculate regressions to 5 performance measures including EPS, Net Income, ROE, ROA and EVA. According to their study the correlations between these performance measures and MVA are: EVA 40%, ROA 13%, ROE 10%, Net income 8% and EPS 6%. The data is from the ten-year period 1986 through 1995.

O'Byrne (1996) from Stern Stewart & Co uses capitalized EVA as independent variable in a regression where market value divided by capital is the dependent variable. He finds that the level of EVA explains 31% of the variance in market value, whereas the level of net operating profit after taxes explains only 17%. When looking at changes in EVA and market value O'Byrne finds that changes in EVA explain 55% of variations in changes in market value. Changes in NOPAT explain only 33%.

Milunovich and Tsuei (1996) review the correlations between MVA and several conventional performance measures in the computer industry. They find EVA to correlate somewhat better with MVA than the other measures. R squared is for EVA 0,42, for EPS growth 0,34 and for ROE and EPS 0,29.

Grant (1996) calculates regression statistics between the MVA-to-capital and EVA-to-capital ratios from the data of 983 firms. He finds explanatory levels (R squared) of 32% with statistical significance. Regressing MVA-to-capital and the spread between return and cost of capital reveals R squared of 37%.

Dodd and Chen (1996) study the correlation between stock returns and different profitability measures including EVA, non-adjusted residual income, ROA, EPS and ROE. In their study ROA explained stock returns best with R squared of 24,5%. The R squared for other metrics are: EVA 20,2%, residual income 19,4% and EPS, ROE approximately 5-7%. The writers concluded that firms adopting EVA might as adopt simple residual income concept, while residual income correlates with share prices almost as well as its adjusted version called EVA. The study is based on 566 U.S. companies from 1983-1992.

Biddle. Bowen and Wallace (1996) present evidence on the relative and incremental information content of EVA, residual income, earnings and operating cash flow. According to the abstract of the study, the writers conclude that "residual income and/or EVA add incremental information in some settings, but that, on average, neither dominates earnings as a performance measure".

Telaranta's study on Finnish Stock markets

The only public study about the correlation of EVA and share prices that has been done on Finnish data is from Tero Telaranta 1997. The study and article based to it concluded that EVA is not any better than traditional performance measures. Many Finnish corporate managers have taken these conclusions very seriously and therefore it is more than justified in this context to examine the study more thoroughly that the studies above.

Telaranta (1997a) study how residual income variables explain movements in market valuations of Finnish companies. The data consist of 42 Finnish industrial companies during 1988 -1995. Only 26 of the companies were listed the whole period and 16 were listed for shorter period. During the research period both the aggregate Market value added and the non-weighted average return on stock among the sample companies are negative. That is because the whole Finnish economy and stock markets experienced a severe recession in the middle of research period around the turn of the decade (1990). Telaranta (1997a) use various different methods in assessing the ability of different measures to explain market movements. As dependent variables he use MVA, market-to-book ratio and excess return on stock. As independent variables Telaranta use two versions of economic profit (residual income) and three versions of Eduard-Bell-Ohlson -figure (near residual income) as well as traditional accounting based performance measures like EBITDA, Operating profit, NOPAT, Net earnings and Cash flow. These all measures are regressed also as percentages of sales and as percentage returns on capital, although using residual income variables in that way is not necessarily theoretically sound. The reason for this is probably to get some comparison material for measures like ROI, ROE, Operating profit % and Net earnings %.

Telaranta's results (1997a) indicate the level of Economic profit (the nearest measure to EVA of all those variables that Telaranta use) to explain 30,7% of the level of Market value added as the next best measure NOPAT explain 30,16%. When talking about changes instead of absolute levels, Economic Profit is the best with R squared of 17,18% whereas Operating profit is the second best with R squared of 16,64%. In several other regressions residual income variables are generally found to be the best measures although with a tiny difference compared to some accounting based variables. In some regressions some accounting based variable is even found to be slightly better than Economic Profit, but these regressions are not very meaningful for one of the following two reasons:

The overall explanatory level with these regressions is far below 5%.

These regressions are on those variables that are all expressed as percentage of sales (e.g. Economic Profit divided with turnover). Economic Profit looses its meaning when expressed as percentage of sales i.e. there is no theory suggesting that variable "Economic Profit/Turnover" should correlate with share prices. Hence there is no meaning attached to the use of Economic Profit with these regressions. Furthermore the explanatory level with these regressions is under 10%.

Telaranta concludes his results to indicate that residual income variables are found to explain the movements in market capitalization with statistical significance. He however founds the explanatory level to be quite low. Telaranta also presents that residual income variables are not found to explain stock returns statistically significantly better than accounting based measures.

Telaranta's results indicate that Economic Profit is the best variable in the study explaining market movements, but that the difference compared to other measures is insignificant. The difference to accounting based measures is naturally low in terms of statistical significance because the data consists of such a small number of firms. The previous studies on the subject have on average about 15-fold number of companies, so the statistical significance between residual income and accounting based measures is also respectively easier to achieve.

Telaranta also states that the overall explanatory lever is lower than in previous studies. On the other hand Telaranta found an explanatory level of 30% (Economic Profit explains MVA) which settles moderately with other studies (Uyemura et al, EVA 40%), (O'Byrne, EVA 31%), (Milunovich and Tsuei, EVA 42%), (Grant, EVA 32%) and (Dodd and Chen, ROA 24,5%; EVA 20,2%; residual income 19,4%). Telaranta achieve also far lower explanatory levels in regressions with Economic Profit as percentage of sales and Economic Profit as percentage returns on capital. These regressions are however not comparable because they are not based on any theory.

Everyone should also notice the effects of the research period on the results. The aggregate Market value added and the non-weighted average return on stock among the sample companies are negative during the period because of the recession. On the other hand Stewart (1990, p.217) has emphasized that possibility of liquidation sets a floor on company's MVA. Telaranta's research period causes thus major bias against EVA. Another unmentioned bias is the use of HEX-index. Especially in the latter half of the period Nokia's and couple of other companies' stocks have a very big weight in HEX-index (Nokia currently7 about 35%). Therefore the impact of one single company is very large in Telaranta's results. If Nokia's stock performance does not correlate very good with Nokia's EVA it would certainly affect the results. Well, Nokia increased its market value to more than 10-fold compared to the beginning of the period and that was of course due to profitable growth prospects (growth in EVA) that the company had ahead. Therefore Nokia's EVA during 1988-1995 hardly explains very much of the change in Nokia's market value during the same period. Nokia´s turnaround from big wealth destroying conglomerate into a big, dynamic wealth creating telecommunications company was firstly seen in share prices and not in profitability measures since investors are staring at future profit prospects (EVA) above the prevailing performance.

Telaranta (1997b) summarize some of his results in Finnish management journal "Talouselämä" in September 1997 and argue that EVA does not create any value added applied as measure or bonus base in companies. Telaranta presents in the article only the results from the regressions where Economic Profit is divided with sales and capital. These regressions do not rank EVA best like the other regressions do. As presented above these regressions are theoretically not sound since any theory does not suggest the variables: "EVA/Turnover" and "EVA/Capital" to correlate with share prices. Also the overall explanatory level of the chosen regressions are very week which also casts doubts on the motives of selection. Why are not the regressions on absolute values selected even though they are theoretically and with explanatory levels far better than those presented in Talouselämä? Telaranta's article has gained a lot of criticism afterwards (Kurikka 1997 from University of Technology, Torppa&Lumijärvi 1997 from KPMG Management Consulting, Lukka&Tuomela 1997 from Turku School of Economics and Martikainen & Kallunki 1997 from University of Vaasa). Main points of this criticism are:

Periodic EVAs can not explain changes in market values caused by changes in long term EVA (Martikainen&Kallunki 1997 and Torppa&Lumijärvi 1997).

Telaranta can not criticize EVA to be week in corporate control and bonus systems, while he has not studied it (Lukka&Tuomela 1997).

General about the correlation of EVA and share prices

The criticism on Telaranta's study mentions at least one fundamental hindrance in estimating EVA-theory with stock price correlations: Market values are above all based on expectations about the future cash flows. Changes in the current share prices thus reflect changes in future cash flow and future EVA expectations. Therefore current EVA can never explain current share prices very well. Change in current EVA might imply some change in future EVA and therefore EVA has some explanatory power. On the other hand the change in future EVA is surely visible also in other measures than EVA. Therefore it is understandable that the other measures have almost as much explanatory power and it is also understandable that the explanatory level is quite low with every measure. Still, current research on the subject seems to suggest that EVA has some additional information compared to conventional measures. However EVA should not be viewed as a magic wand, which can explain current share prices with current performance. The power of EVA is elsewhere, in the field of corporate control, and the rest of this study tries to illuminate it. However if EVA is used as Discounted cash flow to estimate current valuations with future EVA estimates it might be quite informative. Perhaps this is why CS First Boston has trained its research staff in EVA analysis, and Goldman Sachs is about to introduce EVA as "a power tool in the analytical tool kit", as global research chief Steve Einhorn from Goldman Sachs put it (Topkis, 1996, p.265).

Of course the relationship between EVA and MVA can and also has to be tested empirically, but the best way to execute these tests is not to correlate periodic EVA and periodic MVA. One way to assess this theory is to calculate MVAs for certain year (-s) and compare them with EVAs from that year on. That is also the way Stewart does in his study. The problem will still be that MVA accounts for all future EVA and not just for EVA of certain period. The shorter period we take the bigger mistake we make in scope. On the other hand the longer period we take, the worse investors' (EVA) expectations and reality correspond each other. If we compare MVA in 1980 and EVAs in 1981-1990, we assume that investors know in 1980 what is company's EVA in 1981-1990. In the real life investors do not have crystal ball of ten years. EVA critics should construct their studies to test the EVA theory (MVA is discounted EVA) and not purely periodic correlation with share prices. According to the theory, EVA corresponds MVA and not share prices. That is because simply pouring more money in the company can raise share prices. However EVA and MVA do not rise unless that incremental money earns more than its cost of capital. Therefore e.g. EPS and NOPAT capture much better the share price impacts8 of NPV negative investments than EVA. Tests should also take into account the liquidation floor of the value of company, because it is part of the EVA-theory Stewart presents. Thirdly and above all, EVA critics should present some logical and theoretical arguments against EVA. There is no sense making hasty conclusions on the grounds of empirical tests if there is no single logical argument along. Investors have always been interested in return and risk and EVA measures these vital things theoretically be