Introduction

The long-term growth of any business enterprise hinges on its ability to create value for its shareholders. Consequently, the external as well as the internal stakeholders need to know whether value is being created or destroyed, not directly reflected in the audited financial statements of the corporates (the income statement does not account for the charge on shareholders’ funds). In view of the major limitation of the income statement, the study explores the significance of economic value added (EVA) as an optional financial performance measure vis-a-vis conventional earnings measures like earnings after taxes (EAT), net operating profit after tax (NOPAT), earnings per share (EPS), return on equity (ROE) and return on capital employed (ROCE), in shareholders wealth creation. On the one hand, EVA, as a performance benchmarking tool, arms the internal stakeholders with the flexibility in measuring and managing the business risks defined in terms of value creation, a key to success and growth (Sushil 2017); on the other hand, the external stakeholders find an absolute and direct measure of the value created in EVA.

Over the last two decades, value-based performance measures, particularly EVA, are challenging the dominance of conventional earnings measures (Stewart 1991; Young 1997; Banerjee 1999, 2000; Pettit 2000). Stern Stewart & Co., owner of the trademark EVA, contends EVA to be the most effective performance metric that unambiguously measures the shareholders’ value creation. In their view, the true value of the firm is distorted as the accounting practices offer scope for earnings management besides not making provision for cost of shareholders’ funds. Several empirical studies (Biddle et al. 1997; Chen and Dodd 1997, 2001; Ismail 2006; Visaltanachoti et al. 2008; Dimitrios et al. 2009; Kaur and Narang 2010; Bordeianu et al. 2011; Kumar and Sharma 2011; Sharma and Kumar 2012; Altaf 2016) conducted to test the superiority of EVA over conventional financial performance metrics in capturing the intrinsic value of the firm have not found any conclusive evidence in favour of EVA. In contrast, Parvaei and Farhadi (2013), Awan et al. (2014), Ahmed (2015) observe EVA to be superior to conventional financial performance measures in explaining value creation.

In view of diverse and debatable outcomes, the proposed study intends to advance the research on the potential of EVA, a value-based performance benchmark slated to be superior to conventional earnings measures, with a focus on two Asian emerging economies: India and China. The rationale for comparing India with China is manifold; to highlight a few, China and India are the fastest growing economies under BRIC (Brazil, Russia, India and China) countries with huge population and similar demographics (Economy Watch 2010). They are the most favoured investment destinations in Asia–Pacific region owing to their large pool of skilled human resources, infrastructure and a conducive business environment (Lanvin 2017).

Further, these neighbouring countries have secured independence from colonial rules around the same time (India in 1947 and China in 1948), but the economic growth in China has far outpaced that of India (Fig. 1, Table 1). Notwithstanding, both the countries are posing serious challenge to the developed nations.

Fig. 1
figure 1

Data Source: World Bank national accounts data, and OECD national accounts data

Annual GDP growth rate of India and China.

Full size image
Table 1 Annual GDP growth rate of India and China
Full size table

In short, the study empirically compares the value relevance of EVA with ROE, ROCE and EPS, in delivering shareholders’ value for India and China, on a standalone and comparative basis. The study gains significance from the ongoing debate on the potential of EVA as a performance benchmark being greater than that of conventional performance metrics. Secondly, it attempts to gauge the level of value creation by Chinese firms compared to Indian firms, given the fact that Chinese economy has grown faster than the Indian economy. Finally, authors have not come across any such comparative research between India and China.

Literature Review

Performance measurement and management, intrinsic to any business strategy management, have continuously been evolving as separate body of knowledge with a primary focus on financial performance (Sushil 2014). Early nineties witnessed a challenge to conventional earnings measures as appropriate performance benchmarks (Stern et al. 1996). It was primarily on two counts: (a) they do not reflect economic profit, and (b) charge on shareholders’ funds is not accounted for. Stern et al. (1996) identify the spate of hostile takeovers and leverage buyouts of eighties in USA, as capital market response to supposedly inefficient EPS-based model of financial management. Authors’ point out that the EPS-led financial management practices encouraged overinvestment by managers to achieve higher growth. They suggest that the changing dynamics of the fiercely competitive environment warrant an implementation of EVA-based financial management; the system ensures additional investment only if expected returns are more than the total charge on capital employed, and divestment of assets earning inadequate returns. Their contention is supported by Girotra and Yadav (2001) who analyse the benefits and drawbacks of EVA to conclude that EVA motivates managers to think like owners. Effectively, an EVA-aligned performance management offers strategic flexibility by reengineering existing management practices to do away with prevalent ambiguities in shareholders’ value creation, crucial driver for financial performance, continuity and change of any business organisation (Sushil 2015).

The importance of strategic flexibility is also highlighted by Haldar et al. (2016) who observe significant impact of flexibility in corporate governance on international competitiveness. Similarly, Kundi and Sharma (2015) find significant improvement in organisational efficiency by introduction of flexibility in the production process, whereas Shivaani et al. (2015) propose an alternate set of cash flow-based ratios for evaluating investment financing decisions.

Several researches have been conducted to evaluate the efficacy of EVA as a performance benchmark. Chen and Dodd (1997) empirically examine whether EVA can explain the variations in intrinsic value of the firm. Their results show EVA to be useful, but not a perfect performance measure as claimed by Stern Stewart and Co. No more than 26% of variation in stock returns can be explained by EVA putting the relationship on a weak footing. However, EVA performance system fares better than conventional financial measures on count of relative information content. Further, they observe that residual income provides results comparable with EVA. A repeat research (Chen and Dodd 2001) using different methodology to compare value relevance of operating income, residual income and EVA, arrived at the same conclusion. Similar study by Biddle et al. (1997) maintains that in relative information content, EVA does not outperform earnings before extraordinary items.

Clinton and Chen (1998) examine the association of EVA and traditional performance measures with stock prices and stock returns. Their findings indicate an insignificant or even negative correlation of stock prices or stock returns with EVA or residual income. On the other hand, Banerjee (1999) establishes EVA as a significant explanatory variable in an industry-wise relationship study of Indian firms between market value added (MVA) and EVA along with EPS, adjusted return on net worth, capital and labour productivity. However, he refrains from generalising it as a superior performance measure. A follow-up research (Banerjee 2000) examines the relationship between firms’ market value added (MVA) and its current operational value (current EVA) and future growth value (present value of incremental EVAs). The outcome further strengthens the significance of EVA in predicting the market value of the firm, though the dependence is found to be higher on current operational value.

Yet another study on association between EVA and sector returns shows that relative information content of EBIT is better than that of operating cash flows, residual income or EVA (Visaltanachoti et al. 2008). Incremental information content is studied by breaking down EVA into its components, namely operating cash flows, operating accruals, after tax interest expense and capital charge. The test results in operating accruals being the largest contributor in explanation of sector returns. Moreover, information content of operating cash flows outperforms the information contents from after tax interest expense or capital charge. Related research, based on UK companies, supports these outcomes (Ismail 2006). The study observes NOPAT and net income more useful in explaining stock returns compared to EVA and residual income. Another study on Greece firms (Dimitrios et al. 2009) suggests that a pair-wise combination of EVA with EPS significantly increases the explanatory power of stock market returns, though market returns are more closely associated with EPS than EVA.

In contrast, O’Byrne (1996) research claims that changes in EVA over a 10 year horizon explain almost 74% of stock variations compared to around 64% explained by changes in earnings. In a 5-year period, 55% of variations in stock returns can be explained by EVA as against only 24% by changes in earnings. This research further indicates that simple linear regression results in almost identical explanatory power of NOPAT and EVA, but suggests two adjustments required to obtain true picture. Firstly, different EVA multiples for perpetuity value need to be taken for positive and negative EVAs. The reason cited is that market assigns higher multiples for positive EVA. Second adjustment is towards capital multiples that tend to decline with company size. These adjustments significantly enhance the predictive power of EVA over accounting earnings.

On another extreme, Paulo (2002) terms EVA as financial fiction questioning its transmission to market value added (MVA). He states that in an efficient market hypothesis (EMH) world, where required rate of return is equal to internal rate of return, EVA must be equal to zero. So, it is impossible to consistently get abnormal profits. On the other hand, in a non-EMH world, CAPM is not valid, and hence, computation of weighed average cost of capital is not reliable. He also contends that numerous researches have attributed the market movements primarily to factors like volatility, momentum and financial herding instead of earnings. Thus, he challenges the very premise of EVA being the strong driver of stock prices. Chen and Dodd (2002) replyFootnote 1 to Paulo by pointing out that EMH does not prevent firms from earning economic profits till industry equilibrium is achieved in the long run. Secondly, even though CAPM is based on EMH, it continues to dominate market research in the absence of any alternative model. Lastly, they agree that several non-financial factors impact stock prices, but it is the audited financial fundamentals of the firms that are most talked about in capital market.

Kaur and Narang (2010) explore EVA reporting practices prevalent amongst Indian corporates. They examine annual reports of top 500 Indian companies to find that only 37 companies (7.4%) disclose EVA as performance metric. Their study also attempts to identify whether EVA reporting companies are significantly different from EVA non-reporting companies on the basis of their background (age and residential status) and financial performance indicators (size, profitability, leverage, sales efficiency and earnings potential). Findings reveal that the EVA usage and disclosure choice in Indian companies are influenced by company’s size, profitability, leverage, sales efficiency and residential status. The age and earnings potential of a company do not have a significant impact on EVA disclosures choice of a company. Similarly, Pallium (2006) compares the performances of EVA user (75) and non-EVA user (41)—Fortune 500 companies, to assess whether adoption of EVA as performance metric delivers a superior performance. The research notes that EVA-based strategies have not outperformed strategies based on accounting earnings. Besides, the study notes minimal difference in market returns of EVA user and non-EVA user firms. The study actually questions the reliability of EVA as a performance metric since it is observed to be much more unstable relative to the traditional ROI. In a contrarian report, Stewart et al. (2002) show that their EVA clients outperform the market and their peers.

Ivanov et al. (2014) explore portfolio investment strategies for US firms with negative EVA, examining whether negative EVA firm will perform better in future. For the purpose, four groups were formed with the most negative to the least negative EVA firms. Returns on portfolio studied between 2004 and 2009 exhibited abnormal portfolio returns for two groups—least negative and second least negative EVA firms. In addition, firms with low leverage ratios contributed to higher portfolio returns. The research suggests that investors may invest in firms with “near zero EVA” and low leverage to earn abnormal portfolio returns. In another research, Sharma and Kumar (2012) observe that none of the financial performance measures significantly explains stock market variation though positively correlated. Non-financial measures like technology, R&D and quality of management play a prominent role in firm valuation. Therefore, besides capital structure, a large proportion of firm’s valuation is derived from non-financial measures.

A study from China (Wang et al. 2015) investigates the influence of earnings management related to capital cost on EVA for Chinese firms and suggest a method for determining true value of the enterprise also. The study establishes a significant positive association between earnings managed and unadjusted EVA that turns to inverse relationship with adjusted EVA, thereby establishing how earnings management distorts the true value created. Shan (2015) provides further testimony to the fact that the negative impact of value relevance is larger for Chinese firms engaged in earnings management, compared to the ones who exhibit superior corporate governance and restrain in earnings management.

In the year 2010, the State-owned Asset Supervision and Administration Commission (SASAC) of China introduced EVA-aligned executive performance management system, for Central State-owned Enterprises (CSOEs) with an aim to improve their efficiency. Shen et al. (2015) compare the performances of CSOEs pre- and post-implementation of EVA-aligned performance policy with a focus on value of cash holdings. They observe a positive influence of EVA-based performance policy on corporate performance coupled with significant increase in value of cash holdings. Moreover, they assert that this improvement in value is derived from exercising restraint on overinvestment and modification in underinvestment; EVA sets the trend towards efficient use of funds. On the other hand, Chang and Chen (2014) believe that most of the prevalent performance benchmarks, including EVA, are derived from operating profit ignoring the risk of liquidity management. Therefore, in their opinion the introduction of EVA-based performance system for CSOEs by SASAC may not necessarily address the risk of liquidity crunch, given that the cash flow is the lifeline of any firm. Thus, Chang and Chen (2014) posit a modified EVA, as a comprehensive performance metric to incorporate liquidity risk management.

Finally, the literature does not provide any conclusive evidence in favour of EVA as a superior performance benchmark, albeit majority researchers acknowledge its usefulness in value creation. Consequently, a research exploring the utility and superiority of EVA as a performance benchmark for India and China is undertaken.

Research Methodology

Research Objectives and Hypotheses

The purpose of the study is to compare the contributions of economic and financial performance metrics in creating shareholders’ wealth measured by MVA. Several studies have used this measure as a proxy for wealth creation including Young (1997), Banerjee (1999, 2000), Fatemi et al. (2003), Sharma and Kumar (2012), Altaf (2016). Accordingly, the relative and incremental information contribution of EVA against conventional measures is tested using regression analysis (Biddle et al. 1997; Chen and Dodd 1997; Dimitrios et al. 2009; Sharma and Kumar 2012).

Thus, the research aims to empirically examine the information content (relative and incremental) of EVA, ROE, ROCE and EPS, in explaining the variations in MVA. In line with the stated objective, the alternative research hypotheses are formulated as under:

H1

Information content of EVA supersedes that of ROCE, ROE and EPS, in explaining variations in MVA.

H2

The explanatory power of the conventional measures (ROCE, ROE and EPS) in explaining variations in MVA increases significantly in conjunction with EVA.

A set of 260 Indian non-finance companies from NIFTY500 index of National Stock Exchange, India (NSE), form one sample; another set of 254 Chinese non-finance companies from SSE380 index of Shanghai Stock Exchange (SSE) constitute the second sample. Rationales behind selection of these indices are:

  1. (a)

    500 companies under NIFTY500 reflect approximately 94% of the free float market capitalisation of companies listed on NSE, as on 31 March 2016.

  2. (b)

    380 companies under SSE380 comprehensively reflect the performance of the Shanghai new blue chip stocks. Besides, this is the only index with the largest number of companies (380) listed on Shanghai Stock Exchange.

Data spanning 15 years, from 2002 to 2016, have been sourced from ACE Equity database for Indian firms and from Bloomberg for Chinese firms. The final sample (260 Indian and 254 Chinese firms) consists of only those non-finance companies for which the data are available for the entire period from 2002 to 2016.

Definition and Computation of Variables

For performing panel regression analysis, the dependent variable considered is MVA and the independent variables are EVA, ROE, ROCE and EPS. Three variables, MVA, EVA and EPS, are absolute measures, whereas other variables are in ratio form. Therefore, in order to reduce the heteroscedasticity in the data, these three variables are scaled by book value of the total capital employed at the beginning of the year (CEBY) (Fatemi et al. 2003). The logic is that the value is created by the capital invested at the beginning of the year.

MVA

MVA measures the difference in market value of the firm’s securities and the accounting value of the total capital employed, inclusive of debt and preference capital.

$${\text{MVA}} = {\text{Market}}\,{\text{Value}}\,{\text{of}}\,{\text{the}}\,{\text{Firm}}{-}{\text{Book}}\,{\text{Value}}\,{\text{of}}\,{\text{Total}}\,{\text{Capital}}\,{\text{Employed}}$$
(1)

where,

$${\text{Market}}\,{\text{Value}}\,{\text{of}}\,{\text{the}}\,{\text{firm}} = \left( {{\text{Total}}\,{\text{number}}\,{\text{of}}\,{\text{Equity}}\,{\text{Shares}} \times {\text{Price}}\,{\text{per}}\,{\text{Share}}} \right) + {\text{Market}}\,{\text{Value}}\,{\text{of}}\,{\text{Preference}}\,{\text{Capital}} + {\text{Market}}\,{\text{Value}}\,{\text{of}}\,{\text{Interest}}\,{\text{Bearing}}\,{\text{Debt}}$$

and

$${\text{Book}}\,{\text{Value}}\,{\text{of}}\,{\text{Total}}\,{\text{Capital}}\,{\text{Employed}} = {\text{Equity}}\,{\text{Funds}} + {\text{Preference}}\,{\text{Capital}} + {\text{Total}}\,{\text{Interest - Bearing}}\,{\text{Debt}}$$

where,

$${\text{Equity}}\,{\text{Funds}} = {\text{Equity}}\,{\text{Share}}\,{\text{Capital}} + {\text{Reserves}}\left( {{\text{net}}\,{\text{of}}\,{\text{revaluation}}\,{\text{reserves}}} \right)$$

Since the stock price of the company reflects the expected economic value addition over a long-term horizon by the investors, market value of the firm can also be perceived as discounted value of future EVA (Banerjee 2000; Bordeianu et al. 2011). Since MVA (Eq. 1) is a cumulative measure, while all other variables are for the year, incremental MVA over previous year scaled by CEBY is considered for calculation. Scaled MVA (SMVA) is defined as follows:

$${\text{SMVA}}_{\text{it}} = ({\text{MVA}}_{\text{it}} - {\text{MVA}}_{it - 1} )/{\text{CEBY}}_{\text{it}}$$
(2)

The present research assumes that the market value of debt or a hybrid instrument is equal to its book value. Ideally, the market value of a debt or a hybrid instrument is subject to change in prevailing interest rates. But, majority of Indian corporates tend to hold their debt to maturity, so the mark to market value becomes irrelevant. Moreover, the Indian fixed income market is not that mature making it difficult to get the relevant data.

EVA

The estimation of EVA, as per literature surveyed, is non-standard; we have defined it as follows:

$${\text{EVA}} = {\text{Profit}}\,{\text{After}}\,{\text{Tax}} + {\text{Tax}}\,{\text{Adjusted}}\,{\text{Extraordinary}}\,{\text{Expenses}} - {\text{Tax}}\,{\text{Adjusted}}\,{\text{Extraordinary}}\,{\text{Income}} - {\text{Charge}}\,{\text{on}}\,{\text{Shareholders'}}\,{\text{Funds}}\left( {{\text{inclusive}}\,{\text{of}}\,{\text{Prference}}\,{\text{Capital}}} \right)$$
(3)

Scaled EVA (SEVA) is computed as follows:

$${\text{SEVA}} _{\text{it}} = {\text{EVA}}_{\text{it}} / {\text{CEBY}}_{\text{it}}$$
(4)

Economic profit (EVA) is derived by reducing the charge on shareholder’s fund (inclusive of preference capital) from accounting profit (PAT), while making adjustments for extraordinary income or expenses. The rationales behind adjustments for extraordinary items are that they do not form part of company’s regular operations and are not of long-term in nature.

Cost of Equity

Cost of equity (k e ), central to computation of EVA, is another grey area. The study uses Capital Asset Pricing Model (CAPM) to calculate cost of equity (Sharpe 1964). Although the assumptions behind the model are debatable (Paulo 2002), it still remains the most frequently used model for computing cost of equity (Chen and Dodd 1997; Banerjee 1999, 2000; Dimitrios et al. 2009; Ivanov et al. 2014). Thus, the cost of equity is computed as per Eq. 8.

$$k_{e} = R_{\text{f}} + \beta \times \left( {R_{m} - R_{\text{f}} } \right)$$
(5)

where

$$R_{\text{f}} = {\text{Risk-free}}\,{\text{rate}}\,{\text{of}}\,{\text{return}}\,{\text{equivalent}}\,{\text{to}}\,10{\text{-year}}\,{\text{Government}}\,{\text{Bond}}\,{\text{yield }}$$
$$\beta = {\text{Slope}}\,{\text{of}}\,R_{i,t } \,{\text{of}}\,{\text{scrip}}\& {\text{NIFTY}}500,\,{\text{a}}\,{\text{measure}}\,{\text{of}}\,{\text{volatility}}\,{\text{of}}\,{\text{scrip}}\,{\text{against}}\,{\text{market}}\,{\text{movement}}$$

where;

$$R_{it } = {\text{Log}} \left( {P_{it } /P_{it - 1 } } \right)$$
$$R_{m} = {\text{Annual}}\,{\text{market}}\,{\text{returns}}\,{\text{defined}}\,{\text{as}}\,{\text{sum}}\,{\text{of}}\,{\text{price}}\,{\text{retuns}}\,{\text{and}}\,{\text{dividend}}\,{\text{yield }}$$
$$R_{m} = R_{{t\,{\text{index}} }} + \left( {D_{{t\,{\text{index}} }} /P_{{t - 1\,{\text{index}}}} } \right)$$

Since equity is a long-term investment, the risk-free rate of return is benchmarked to 10-year Government bond yield as a measure of long-term returns. Monthly closing price of each scrip and the market index (NIFTY500) is taken to estimate β. A period of 5 year is selected to reduce the impact of market volatility. Similarly, \(R_{m}\) is computed as the mean of prior 60 monthly log returns of index price scaled to annual value, plus the average dividend yield of corresponding 5 years.

Return on Equity (ROE)

The ratio is chosen for the purpose of comparison, as it is closest to measuring value creation for equity shareholders, under conventional financial measure. ROE, a measure of profitability against shareholders’ investments, is defined as:

$${\text{ROE}} = \frac{{\left( {{\text{Earnings}}\,{\text{after}}\,{\text{Tax}} - {\text{Preference}}\,{\text{Dividend}}} \right)}}{{{\text{Average}}\,{\text{Equity}}\,{\text{Funds}}}} \times 100$$
(6)

where equity funds will include equity share capital plus reserves and surplus of the company, after deducting revaluation reserves and accumulated losses, if any.

Return on Capital Employed (ROCE)

This ratio estimates the returns earned by the firm on its long-term capital employed.

$${\text{ROCE}} = \frac{{\left( {{\text{Earnings}}\,{\text{after}}\,{\text{Tax}} + {\text{Interest}}\,{\text{net}}\,{\text{of}}\,{\text{tax}}\,{\text{advantage}}} \right)}}{{{\text{Average}}\,{\text{Capital}}\,{\text{Employed}}}} \times 100$$
(7)

Theoretically, capital employed would include only long-term sources of finance, namely shareholders’ funds plus long-term debt. But, we have included—the short-term interest-bearing debt for two reasons. First, the data available on interest expense are not segregated into interest on long-term and short-term debts. Second, the large amount of short-term bank credit enjoyed by Indian firms is mostly rolled over year on year, effectively assuming the characteristics of long-term debt.

This ratio deals with accounting profitability on total capital employed, akin to calculation of charge on total capital employed while computing economic profitability.

Earnings Per Share (EPS)

It is the net income per share available to equity investors after deducting preference dividend.

$${\text{EPS}} = \frac{{\left( {{\text{Earnings}}\,{\text{after}}\,{\text{tax}} - {\text{Preference}}\,{\text{Dividend}}} \right)}}{{{\text{Number}}\,{\text{of}}\,{\text{ourstanding}}\,{\text{equity}}\,{\text{shares}}}}$$
(8)

It is a widely used ratio, especially in capital market to assess corporate performance. Studies have shown that stock market returns of a firm are closely associated with EPS (Dimitrios et al. 2009) making it a suitable measure for comparison. Since EPS is an absolute measure, it is scaled for regression analysis. Standardised EPS (SEPS) is defined as per Eq. (9):

$${\text{SEPS}} _{\text{it}} = {\text{EPS}}_{\text{it}} /{\text{CEBY}}_{\text{it}}$$
(9)

Growth Variables

If the current performance metrics (EVA, ROE, ROCE and EPS) assess the present financial health of the firm, the growth of these variables over previous year conveys the growth prospects of the firm. Accordingly, the study includes ΔEVA, ΔROE, ΔROCE and ΔEPS as independent variables, defined as percentage growth over previous year.

Testing of Hypotheses

The following models, adopted by a number of previous studies (Easton and Harris 1991; Biddle et al. 1997; Chen and Dodd 2001; Dimitrios et al. 2009; Sharma and Kumar 2012), have been used to test the hypotheses related to relative and incremental information content. The model assumes a linear relationship amongst variables, zeroing on panel regression analysis as the research tool. The hypothesis for relative information content (H1) has been tested using the following regression models:

$${\text{SMVA}}_{\text{it}} = a_{0} + a_{1} {\text{SEVA}}_{\text{it}} + a_{2} \Delta {\text{EVA}}_{\text{it}} + u_{1}$$
(10)
$${\text{SMVA}}_{\text{it}} = b_{0} + b_{1} {\text{ROCE}}_{\text{it}} + b_{2} \Delta {\text{ROCE}}_{\text{it}} + u_{2}$$
(11)
$${\text{SMVA}}_{\text{it}} = c_{0} + c_{1} {\text{ROE}}_{\text{it}} + c_{2} \Delta {\text{ROE}}_{\text{it}} + u_{3}$$
(12)
$${\text{SMVA}}_{\text{it}} = d_{0} + d_{1} {\text{SEPS}}_{\text{it}} + d_{2} \Delta {\text{EPS}}_{\text{it}} + u_{4}$$
(13)

The relative significance of EVA and conventional measures is evaluated by comparing their coefficients (β) and explanatory powers (R 2), derived from regressing each of the above equations.

Incremental information content (H2) has been tested by conducting regression analysis of MVA with pair-wise combination of EVA and one of the financial measures (Chen and Dodd 1997, 2001; Dimitrios et al. 2009).

$${\text{SMVA}}_{\text{it}} = \varUpsilon_{0} + a_{1} {\text{SEVA}}_{\text{it}} + a_{2} \Delta {\text{EVA}}_{\text{it}} + b_{1} {\text{ROE}}_{\text{it}} + b_{2} \Delta {\text{ROE}}_{\text{it}} + u_{5}$$
(14)
$${\text{SMVA}}_{\text{it}} = \varPhi_{0} + a_{1} {\text{SEVA}}_{\text{it}} + a_{2} \Delta {\text{EVA}}_{\text{it}} + c_{1} {\text{ROCE}}_{\text{it}} + c_{2} \Delta {\text{ROCE}}_{\text{it}} + u_{6}$$
(15)
$${\text{SMVA}}_{\text{it}} = \alpha_{0} + a_{1} {\text{SEVA}}_{\text{it}} + a_{2} \Delta {\text{EVA}}_{\text{it}} + d_{1} {\text{SEPS}}_{\text{it}} + d_{2} \Delta {\text{EPS}}_{\text{it}} + u_{7}$$
(16)
$${\text{SMVA}}_{\text{it}} = V_{0} + a_{1} {\text{SEVA}}_{\text{it}} + b_{1} {\text{ROE}}_{\text{it}} + c_{1} {\text{ROCE}}_{\text{it}} + d_{1} {\text{SEPS}}_{\text{it}} + u_{8}$$
(17)

The incremental value of pairing EVA with ROE has been obtained by reducing adj. R 2 value of Eq. 12, from Eq. 14. The resultant incremental adj. R 2 represents the incremental information available in explaining variations in MVA on combining EVA with ROE (Sharma and Kumar 2012; Dimitrios et al. 2009). On the same lines, incremental adj. R 2 values for other pair-wise combinations of variables have been obtained, EVA with ROCE (Eqs. 15–11) and EVA with EPS (Eqs. 16–13). Lastly, Eq. 17 provides the level of explanatory power of all the performance measures put together. Therefore, its incremental information content is arrived at by reducing the adj. R 2 generated by regressing this equation without EVA.

Unit root tests performed using Levin–Lin–Chu and Hadri LM methods reveal that the sample data are stationary, ensuring validity of the standard assumptions for asymptotic analysis. In order to test the predictability of the regression models, the total time period (2002–2016) of the study has been split into sub-periods, an in-sample period of 2002–2011 for parameters estimation and an out-of-sample period of 2012–2016 for forecasting using the estimated coefficients. The goodness of fit is then tested by comparing the means of actual and forecasted values of MVA during out-of-sample period, using T test.

Empirical Results and Discussion

India

Descriptive statistics of all the variables and Pearson’s correlation are tabulated in Table 2. The outcome exhibits weak correlations between MVA and the performance metrics selected for the study, with the highest correlation coefficient of ROCE at 0.23. Moreover, the correlation coefficient of EVA at 0.16 is only marginally higher than that of ROE at 0.14, with EPS sharing the bottom space at 0.10.

Table 2 Descriptive statistics and Pearson’s correlation, India
Full size table

The results are at odds with the findings of Chen and Dodd (1997) and Sharma and Kumar (2012), which depict a strong correlation between EVA and MVA valued at 0.45. However, it is similar to the results of Ismail (2006), where coefficient of correlations of EVA with stock returns is only 0.11. It is interesting to note that the outcome of Sharma and Kumar (2012) puts EPS and ROCE at par with a correlation coefficient of 0.17, whereas ROE is the lowest at 0.04.

In contrast, we find EPS to exhibit the least correlation at 0.10, with ROCE at the top with 0.23. Moreover, ROE is also reasonably high and closer to EVA at 0.14. The result does not support the contention of the proponents of EVA, assigning it to be highly associated with value creation. Having evidenced the presence of linear relationship between MVA and the four performance measures, the next step is to define the causal relationship using regression analysis.

Relative Information Content

The outcome of panel regressions for relative information content (Eqs. 10, 11, 12 and 13) is detailed in Table 3. The pooled OLS regression method is identified to be the most suited model for the panel data under study; the post-estimation tests, F-test for fixed effects and Breusch and Pagan Lagrangian multiplier test for random effects, do not find any significant presence of fixed or random effects across models.

Table 3 Regression results for relative information content, India
Full size table

The outcome indicates weak explanatory powers of all the performance measures with adj. R 2 values ranging between 5.3% (ROCE) and 0.8% (EPS). ROE and EVA are pegged at par with the ability to explain only 2.6% (adj. R 2) of the variation in MVA. The R 2 values of conventional performance measures are similar to the outcomes of Kumar and Sharma (2011)—ROCE (4%), ROE (6.6%); Dimitrios et al. (2009) from Greece, another emerging market—ROCE (4%), ROE (Nil) and EPS (1.9%). But, Kumar and Sharma (2011) find R 2 for EVA to be 14%, although our result aligns itself with Dimitrios et al. (2009) findings for EVA (0.9%). In another Indian study, Altaf (2016) finds the value of R 2 of EVA for manufacturing sector at 9.2% and service sector at 17.4%. Notwithstanding, the relatively higher explanatory power of EVA, akin to our result, both the Indian studies rank value relevance of EVA lower than that of conventional performance measures like operating income and cash flows. Nevertheless, we find that the coefficients of all the current performance measures are significant at 0.001 level of significance. In addition, the models are able to forecast MVA in the out-of-sample period test (p > 0.05). We have not dropped the growth variables from the equations, in spite of their not reflecting significant relationship with MVA, as they add to the R 2 value.

Given the similar explanatory powers of independent variables, a comparison of coefficients of the current performance variables establishes ROCE to be having the maximum impact on MVA—a unit change in ROCE results in 3.734 units change in MVA. It is closely followed by EVA with coefficient at 2.468, whereas EPS with a coefficient of 1.747 fares marginally better than ROE with coefficient at 1.217. The findings are consistent with the outcome of several studies (Easton and Harris 1991; Biddle et al. 1997; Visaltanachoti et al. 2008; Dimitrios et al. 2009; Kumar and Sharma 2011, Sharma and Kumar 2012) that do not find EVA outperforming conventional performance measures. Pallium (2006) finds minimal difference between stock performances of EVA user and non-EVA user firms, putting a question mark on power of EVA as a tool to deliver superior performance. In contrast, Chen and Dodd (1997) find that the value-based measures (EVA per share, standardised EVA, spread and growth) explain 41.5% variations in stock returns as compared to 36.5% explanation offered by conventional financial measures (ROA, ROE and EPS). Moreover, Banerjee (2000) establishes a significant relationship between EVA, ΔEVA and MVA for Indian firms, with an adj. R 2 of 74%.

In sum, ROCE exhibits the largest impact on MVA, although closely followed by EVA.

Incremental Information Content

Regression analysis for incremental information content is tabled in Table 4. All the independent variables tested for multicollinearity using variance inflation factor (VIF) do not exhibit any significant multicollinearity; the values of VIF across models (Table 4) are less than five (Neter et al. 1985). Moreover, the F tests and LM tests result in favour of pooled regression technique as against fixed or random effect methods, for all the models.

Table 4 Regression results for incremental information content, India
Full size table

The results indicate that the pair-wise combination of EVA with EPS offers a maximum increase in adj. R 2 of 1.9%, unlike the results of Sharma and Kumar (2012) where maximum incremental R 2 of 2.3% is registered by EVA in conjunction with ROCE. However, ROCE in conjunction with EVA renders the impact of EVA on MVA negative and insignificant (Eq. 15: a 1 = − 0.122, p = 0.770). Simultaneously, ROCE registers a marginal increase in its coefficient (Eq. 15: c 1 = 3.904, p = 0.000), compared to its univariate regression (Eq. 11: c 1 = 3.734, p = 0.000). Moreover, on regressing all the performance measures taken together, only the coefficient of ROCE is significant (Eq. 17: c 1 = 3.883, p = 0.000). Basically, the combination of ROCE with other performance measures (Eqs. 15 and 17) provides a result similar to its reduced form of univariate regression (Eq. 12), with similar adjusted R 2 value (5.3–5.4%). Moreover, the incremental adj. R 2 for the combinations of EVA with ROE is negligible at 0.6%.

The outcome is in contradiction to the results of multivariate regression analysis of Altaf (2016), Kumar and Sharma (2011) and Sharma and Kumar (2012) where the coefficient of EVA displays positive and significant relationship with MVA, albeit, inferior to operating income or ROCE in relative terms. However, it is important to note that all the four regression models for incremental information (Eqs. 14, 15, 16 and 17) can successfully forecast MVA (Table 4: T test for predictability). Consequently, we fail to defend H2 that the combination of EVA with conventional performance measures offers significant incremental information in explaining variation in MVA. The findings are aligned with the observations of Biddle et al. (1997) showing negligible incremental information content on pairing value-based performance measures with traditional performance indicators. Contrary to our results, Chen and Dodd (1997) ascertain an incremental R 2 of 10.5% on combining EVA with traditional performance variables. Dimitrios et al. (2009) find a substantial increase in R 2, from 1.9 to 7.2%, on combining EVA with EPS.

In effect, the present research does not support the hypothesis that the information content of EVA supersedes that of ROE, ROCE and EPS, in explaining variations in MVA. Moreover, no significant incremental information is derived from EVA in conjunction with conventional performance measures. Yet, the relative significance of EVA in predicting MVA of a firm is next only to ROCE, corroborating the value relevance of EVA as financial performance measure.

China

Pearson’s product moment correlation coefficients along with the descriptive statistics are presented in Table 5. Clearly, EVA seems to display a feeble negative correlation with MVA (− 0.03), in contrast to many studies (Sharma and Kumar 2012; Chen and Dodd 1997; Ismail 2006), including the present research where EVA of Indian firms posits a relatively significant and positive correlation (0.16).

Table 5 Descriptive statistics and Pearson’s correlation, China
Full size table

The conventional measures do exhibit positive, but a weak correlation; the extent of correlation is almost similar with EPS (0.17) showing the highest correlation followed by ROCE (0.13) and ROE (0.12).

Relative Information Content

As per the F tests and LM tests, undertaken pooled regression technique is applied for regression. The outcomes for relative information content (Table 6) corroborate the weak correlation result for EVA; regression analysis neither finds EVA (β = − 0.022, p = 0.964, adj. R 2 = 0.00) significant, nor does it offer any explanation to variations in MVA. In case of Chinese firms, EPS (β = 667.7, p = 0.000, adj. R 2 = 3.6%) seems to have huge impact on MVA and offers the highest explanation to variations in MVA, in relative terms. A unit change in SEPS is expected to increase SMVA by 667.7 units. ROCE (β = 4.219, p = 0.000) is the next significant measure with adj. R 2 at 2.7%, followed by ROE (β = 5.540, p = 0.000) with an explanatory power of 1.9% (adj. R 2).

Table 6 Regression results for relative information content, China
Full size table

Thus, EPS establishes to be the leading performance indicator, while EVA is not only insignificant, but negatively correlated.

Incremental Information Content

Table 7 summarises the panel regression output for incremental analysis. The method selection tests (F test and LM test) favour pooled regression technique for all the models. In addition, the multicollinearity levels amongst IVs are also observed to be low (VIF < 5).

Table 7 Regression results for incremental information content, China
Full size table

Clearly, the highest incremental value is derived from Eq. 17 (incr. adj. R 2 = 4.1%, adj. R 2 = 7.4%), while the lowest value addition comes from EVA-EPS model (incr. adj. R 2 = 1.6%, adj. R 2 = 5.3%). Again, both these models are not validated out-of-sample period; therefore, the conclusion cannot be generalised. Hence, the EVA-ROCE model (incr. adj. R 2 = 3.1%, adj. R 2 = 5.8%) proves to offer the maximum incremental and relative explanatory power; the EVA–ROE model (incr. adj. R 2 = 2.5%, adj. R 2 = 4.4%) provides only marginally lower incremental and relative explanatory quotients.

Effectively, EVA does not appear to be an important performance indicator for the Chinese firms. All the conventional measures are significant with EPS displaying the highest impact on MVA in relative terms, but the EVA in conjunction with ROCE delivers the maximum incremental content.

Comparison

A snapshot of relative comparative assessment between India and China is tabulated in Table 8. Evidently, ROCE appears as the most significant performance measure for India against EPS for China.

Table 8 Relative comparison at a Glance—India versus China
Full size table

Apparently, unlike India, Chinese stakeholders rely heavily on EPS for performance assessment; a unit change in SEPS tends to increase SMVA by 667.0 units. Moreover, for India, EVA fares better than ROE or EPS, while the measure seems inconsequential for China. In addition, no significant incremental information is derived from EVA in conjunction with the conventional performance measures for India; an incr. adj. R 2 of 3.1% is delivered by EVA in combination with ROCE for China, even though relatively it posits to be an insignificant performance metric.

Family-owned business contributes to two-thirds of Indian GDP as per KPMG survey of 2013 (Global Data Points, Family Firm Institute). We believe that the plausible reason for EVA to rank almost at par with ROCE for Indian firms is mainly because the owners holding substantial stakes in the firms are also the managers of the firms; EVA is expected to motivate mangers to think like owners. Besides, the Indian GAAP offers limited scope of earnings management that tends to distort the true value of the firms. The case of China of EVA exhibiting negative relationship with MVA is actually inexplicable given the fact that Chinese government has recognised EVA as an executive performance benchmark and the Chinese government holds a minimum stake of 50% in approximately 40% of the SSE380 index firms covered in the study. Moreover, the extensive impact of EPS on MVA reflects that the Chinese capital market continues to operate in a regulated environment. In addition, there is not enough clarity on earnings management practices adopted by Chinese firms (Shan 2015; Wang et al. 2015) to draw any meaningful conclusion.

Concluding Observations

To conclude, the empirical results to test relative and incremental contributions of financial performance measures (EVA, ROE, ROCE and EPS), in explaining variation in shareholders’ wealth creation (MVA as proxy), fail to establish EVA as a superior performance benchmark for both the countries. Thus, the outcomes are akin to the findings of many scholars (Biddle et al. 1997; Pallium 2006; Visaltanachoti et al. 2008; Dimitrios et al. 2009; Kumar and Sharma 2011; Sharma and Kumar 2012; Altaf 2016) that contradict the contention of Stern Stewart & Co., proponents of EVA, claiming it to be a measure superior to conventional performance measures in assessing value creation. Nevertheless, in Indian context, we find a positive and significant relationship between EVA and MVA (β = 2.468, p = 0.000), next only to ROCE (β = 3.734, p = 0.000) which is found to have the maximum impact on MVA amongst the variables under study. Consequently, we believe that the value relevance of EVA cannot be ignored, similar to the assertions of many extant literatures, whether they could establish its superiority or not. Even in the case of China, EVA provides significant incremental information in conjunction with ROCE, in explaining the variation in MVA; therefore, it cannot be summarily ignored though relatively it is observed to be insignificant. Even so, the study highlights the negative relationship of EVA with the value creation by Chinese firms, in spite of the fact that China has witnessed a substantially higher annual GDP growth rate over the period 2002–2014, in comparison with India; the plausible reason it is now observing a slowdown (Fig. 1).

The outcome, particularly from Indian point of view, corroborates our view that EVA has all the potentials of an optimal performance benchmark for Indian corporates. The primary reason for not adopting EVA as a performance benchmark, in our view and supported by the literature, is the complex and non-standard method of its computation. To conclude, the need of the hour is to standardise the computation of EVA and ensure its mandatory disclosure in financial statements as an absolute measure of shareholders’ wealth creation, a significant move towards offering flexibility in performance measurement to internal as well as external stakeholders.

Though the research has produced some significant findings, it is limited in terms of aggregating the firms across sectors and does not include any distinction between EVA user and non-EVA user firms. Therefore, a disaggregative analysis in terms of sectors, ownership, size and age of the companies may offer more depth to the research, as a way forward. Moreover, as the Indian firms move from GAAP to International Financial Reporting Standards (IFRS) for accounting, from the financial year 2017–2018 (PwC India 2015), it widens the scope of research on how the changes in accounting practices may influence the value relevance of EVA and conventional financial performance metrics.