Print Reference This Reddit This

Evidence That Earnings Outperform Historical Cash Flows Finance Essay

Prior studies have provided evidence that earnings outperform historical cash flows in predicting future cash flows. Later researchers reveal that the major accrual components of earnings each possess significant explanatory power in predicting future cash flows and they augment, rather than replace, the predictive ability of aggregate earnings. The degree of relationship of Earning, Sales & Cash Flows with Share Price is the Primary goal of this study though Prior study focused on predictive abilities of Cash Flow, Sales and Earning where as this study also predicts the abilities of Variables by Using share price as the dependent variable and as a proxy for future cash flows. The result provide evidence that Cash Flow does not appear to be significant with share price with the significance value of 0.061 , Sales also appears insignificant with the significance value of 0.108 & Earning appears significant with the significance value of 0.000 and relationship exist between Earning & Share Price.


Earnings occupy an important role in financial accounting research and finance, this emphasis on earnings in the literature exists for many reasons. It is the most widely accepted measure of firm performance.

Attention is also given to earnings because it is commonly used in evaluating management performance. Perhaps the biggest reason for the attraction to earnings, though, lies with the notion that earnings serves as a predictor of future cash flows. Many theories have represented that the accrual earnings represents a better predictor of future cash flows than historical cash flows. A company’s existence depends on its ability to generate positive cash flows, and research has demonstrated that share price is directly related to an entity’s cash flow prospects. Thus, because earning is viewed as a primary predictor of future cash flows, it is also a key determinant of share price.

Recent research (Barth et al., 2001) has shown that the predictive ability of earnings can be improved when it is disaggregated into its major accrual components. One of the components is sales revenue, which surprisingly has been largely ignored in the literature as a predictor of future cash flows and share price. The degree of relationship of earnings and cash Flow, sales with share price is a goal of the study.

Information in the Cash Flow statement help investors, creditors and other users of financial statements to assess attributes Such as the firm’s Liquidity financial flexibility and risk. Financial Statements gives information about firm’s cash flows and users are interested in cash flows mainly as these affect their future cash flows. People are interested and they are concerned with “assessing the entities ability to meet its obligations and paying dividends where as accounting books state that earning rather than Cash flows provides a better indication of an enterprise’s to generate favorable cash flows. Short-term cash flow predictions can also provide information to investors by trend in cash flows. A firm that raising owners equity/ capital & growing and recording large positive accruals relative to assets, where as a firm that is declining or downsizing will be recording also large negative accruals relative to assets. The reason behind the accounting rules that apply to growing and declining or downsizing firms differ fundamentally and this difference likely stems from the historical emphasis on reliability and conservatism in accounting.

A firm’s ability to generate cash flow affects the value of its securities, so the ability to assess future cash flow is important for the investment community, both shareholders and creditors. While shareholders may be concerned with the stream of cash flows to perpetuity, many creditors are concerned solely the short-term cash-generating ability of a company.

The next section of this study provides Literature Review. Section 3 describes the Research question. Section 4 describes Research Method & Data collection. Section 5 describes Results which includes Tables and Section 6 describes conclusion.


This study relates to examine the relationship of cash flow from operations, earning and sales with share price and the previous research has predicted the comparative abilities of cash flow, earning and sales but this study is only concerned with the relationship of cash flow, earning and sales with share price.

In the finance literature that market forces determine share price equal to the discounted value of a stream of expected future cash flows (Hollister et al., 2002). Cash flows represent amounts investors expect to receive in the form of dividend payments or from the sale of their shares and not necessarily the annual operating cash flows generated by a firm. Consequently, it is in a very broad sense that share price is considered to embody a firm’s future cash flows. Even if share price is often thought of and evaluated in terms of cash flows, earnings is also known to be extremely important to managers and analysts because of the key information it conveys about future prospects (Brigham and Ehrhardt, 2002).

Richardson et al. (2006) provided new evidence that indicated the temporary accounting distortions were a significant contributing factor to the lower persistence of the accrual component of earnings and evidence also showed that the lower persistence of accruals extends to accruals that are unrelated to sales growth and that extreme accruals were systematically associated with alleged cases of earnings manipulation.

Dechow and Ge (2006) found that earnings were more persistent than cash flows in high accrual firms, due to the growth of firms. If the firm was growing, allowable accruals adjustments were likely to reduce the effect of negative transitory cash flows on earnings, and consistent with this prediction, high accrual firms have high earnings persistence relative to that of cash flows.

Extending the idea that is developed by Dechow et al. (1998) that the accruals within earnings enable earnings to outperform operating cash flows in predictions, Barth et al. (2001) examined the major accrual components of earnings. The researcher disaggregated earnings into its cash flow component and six major accrual components of earnings (i.e. change in inventory, change in accounts receivable, change in accounts payable, depreciation, amortization, and other accruals). The Barth et al. (2001) premise for disaggregating earnings into these components was that aggregate earnings masks information contained in the individual components and that the each major component of earnings reflects different or unique information about future cash flows. Study results showed that each of the six accrual components was significant in predicting future cash flows and models with earnings disaggregated into the six accrual components and the cash flow component markedly outperformed models developed with aggregate earnings in terms of predicting future cash flows.

Chan, Chan, Jegadeesh, and Lakonishok (2006) investigated whether high accrual

firms have large income decreasing special items in the following year and reseacher found some evidence consistent with this conjecture and interpret this finding as the reversal of earnings management in the prior year. However Researchers do not investigated or addressed the contemporaneous relation between special items and low accrual firms.

Richardson, Sloan, Soliman, and Tuna (2005) found that the less reliable accruals result

in lower earnings persistence and investors do not fully anticipate the lower earnings

persistence. Two accrual categories of low reliability are change in current operating

assets (COA) and change in non current operating assets (NCOA). Change in COA is

dominated by receivables and inventory. Change in NCOA is dominated by PPE and

intangibles. Both changes of COA and NCOA could reflect write downs due to special

items. Researchers do not address whether there is any systematic difference between low and high accrual firms in terms of changes of COA and NCOA, nor do they investigate the role of special items.

Al-Attar and Hussain (2004) extended the research study work of Barth et al. (2001) to examine U.K. entities like Barth et al. (2001), and found that disaggregating earnings into its major accrual components and cash flow component produced models with explanatory power superior to that of aggregate earnings relative to predicting future cash flows.

Jussi Nikkinen and Petri Sahlström (2004) extended the study of cash flow prediction model of Barth et al. (2001), which disaggregating earnings into cash flow and the components of accruals with the year and country effects, and to examine the impact of accounting environment on cash flow prediction. The result performed well in countries where the accruals are used mainly to correct cash flow to better reflect current profitability of the firm, i.e. in countries with high information content of accruals. It implied that the cash flow prediction model by Barth et al. (2001) can be used in different kinds of accounting environments.

Hollister et al. (2002) also expanded the work that done by Barth et al. (2001) by examining companies in the U.S., U.K., Germany, and Japan and included countries other than the U.S. because researcher believed that the accrual components of earnings would be less important in predicting future cash flows for these countries compared to the U.S. This is because earnings in the non-U.S. countries must conform more to reported taxable income or are subject to greater earnings management than in the U.S.

Though Hollister et al. (2002) discovered that, for every country examined, earnings disaggregated into the major accrual components and the cash flow component predicted future cash flows better than either current operating cash flows or aggregate earnings.

Dechow and Dichev (2001) test for the informativeness of working capital and results suggested that accruals correct for the timing problems of cash flows but at the cost of including errors in estimation that lead to lower accrual quality for earnings persistence.

Black (1998) studied the impact of the life cycle stages on firms’ performances over a wide sample of firms. Looking at four life cycle stages (start-up, growth, maturity, and decline), researcher documented that accrual earnings were more value relevant in mature stages, whereas cash flows are more value relevant in stages characterized by growth. Growth industries include those with intangible intensive, high technology characteristics, which suffer from timing and matching problems.

Dechow et al. (1998) examined cash flow and the accrual process related to accounts receivable, accounts payable, & inventory to derive the prediction that current earnings are the best predictor of future cash flow. Study reported that the firm specific variation in cash flow forecast errors based on the aggregate earnings is significantly lower than that based on cash flow and researcher also reported that in firm specific regression of future cash flow on current aggregate earnings & cash flow, both have incremental explanatory power.

Das et al. (1998) examined the association between Value Line analysts' earnings forecasts and earnings predictability and found that as earnings become less predictable analysts' earnings forecasts become increasingly optimistic.

A research line has studied the performance of earnings based valuation relative to discounted cash flow and other discounting methods. The findings (Penman and Sougiannis, 1998, Francis, Olsson and Oswald, 2000) indicated that the over relatively short forecast horizons, ten years or less, valuation estimates using the earnings approach generate more precise estimates of value than discounted cash flow models and this advantage for the earnings based approach persists for firms with conservative or aggressive accounting, indicated that accrual accounting in the U.S. did a reasonably good job of reflecting future cash flows.

Ray, Pieter, May & Lynn (1998) evaluated the relation between security returns and funds based earnings components and researchers documented that proxy for market expectations of the components that are based on measures of historical serial and cross dependencies were substantially more accurate than random walk proxies. However, researchers detect significantly higher valuations of the operating cash flow component of earnings, relative to current accruals, when market expectations represented using the dependency based predictions. Such differential valuation was not detectable for random walk representations.

Burgstahler and Dichev (1997) provided evidence that the quoted firms manage reported earnings to avoid earnings decreases and losses and found unusually low frequencies of small decreases in earnings and small losses and unusually high frequencies of small increases in earnings and small positive income and also found evidence that two components of earnings, cash flow from operations and changes in working capital were used to achieve increases in earnings. If earnings were managed than it could be expected that firms with negative earnings show more negative and volatile cash flows than firms with positive earnings.

Dechow et al. (1995) examined a sample of earnings manipulations subjected to SEC enforcement actions and found that these earnings manipulations were primarily attributable to accruals that reverse in the year following the earnings manipulations. Thus, the evidence was consistent with earnings management contributing to the lower persistence of the accrual component of earnings.

Even though Kaplan (1994) did not examine the relationship between sales and cash flows or share price, showed that sales performance measures, earnings, and stock returns are all key factors in top executive compensation and turnover. Dechow and Dichev (2002) and Francis et al. (2004) note that Earnings quality can be affected by sales volatility (Dechow and Dichev (2002) and Francis et al. (2004). By and large the greater the sales volatility, the more unstable is the operating environment. This results in larger estimation errors for accruals and diminished earnings quality.

Significant research (e.g. see Watts, 1977; Dechow, 1994; Bartov et al., 1997) suggestsed that the earnings reflects a stronger correlation with value (i.e. share returns) than does current operating cash flows. Although the researchers examined the value in terms of share returns, Dechow (1994) notes that the substitution of raw stock prices provides analogous results. Studies have also shown that earnings better predicts future operating cash flows than does current operating cash flows (e.g. see Greenburg et al., 1986; Murdoch and Krause, 1989; Dechow, et al., 1998). The reasoning for this is because accruals in earnings “offset the negative correlation in cash flow changes to produce earnings changes that are much less negatively serially correlated (Dechow, et al., 1998, p. 163).” Dechow et al. (1998) explain that this is why earnings, rather than current operating cash flows, tends to be used in firm (i.e. share) valuations.

Lorek & Willinger (1996) examined the time series properties and predictive abilities of cash flow data , Results indicated that the model clearly outperforms firm specific and common structure ARIMA models as well as a multivariate, cross-sectional regression model popularized in the literature Findings are robust across alternative cash-flow metrics (e.g., levels, per-share, and deflated by total assets) and are consistent with the viewpoint espoused by the FASB that cash-flow prediction is enhanced by consideration of earnings and accrual accounting data.

Finger (1994) found out the earnings ability to predict future earnings and future cash flow from operations1 one through eight years ahead using annual data from1935-87 for 50 firms. In this research time series methods used to test firm-specific predictive ability over the entire time period (hereafter in-sample regression tests) and then compare out of sample forecast errors to assess earnings' ability to improve earnings or cash flow forecasts up to eight years ahead. Thus earnings are a significant predictor of future earnings. The random walk provides better out of sample forecasts than do individually estimated models one year ahead for 52% of the sample firms, Out of sample forecasts show that random walk models outperform individually estimated earnings models for one-year but not for four- or eight-year horizons. Earnings, used alone and with cash flow, are a significant predictor of cash flow for the majority of firms. However out-of-sample forecasts show that adding earnings rarely improves cash flow forecasts. Cash flow is a better short-term predictor of cash flow than are earnings, both in and out of sample, and the two are approximately equivalent long-term.

The nature of the information contained in the accrual and cash flow components of earnings and the extent to which this information is reflected in stock prices (Sloan 1996). It was found that earning performance attributable to the accrual component of earnings exhibits lower persistence than earnings performance attributable to the cash flow component of earnings, hence results also indicated that stock prices act as if investors "Fixate" on earnings, failing to distinguish fully between the different properties of the accrual and cash flow components of earnings.

Juan M. Rivara (1996) found out the accuracy and the consensus among forecasters of earnings estimates for U.S. domestic and U.S. multinational corporations, it was observed that the accuracy of earnings forecasts is significantly lower for purely domestic firms than for U.S based multinationals. Like wise the level of consensus in earnings estimates submitted by financial analysts is significantly lower for U.S. domestic than for U.S. multinational firms.

The accounting profession requires that firms disaggregate net income into specific components, even though earnings disaggregation is important for assessing firm profitability, there is little empirical evidence that the classification scheme actually improves profitability forecasts by analyzing the accuracy improvements in out of sample forecasts of one year ahead return on equity (ROE) to examine the predictive content of earnings disaggregations (Fairfield, Sweeney, & Yohn 1996) .Result show that the classification scheme prescribed by the accounting profession does increase the predictive content of reported earnings. It was found forecasting improvements from earnings disaggregation. These improvements go beyond separating extraordinary items and discontinued operations from the other components of earnings. Further disaggregation of earnings (into operating earnings, non-operating earnings and taxes, and special items) improves forecasts of ROE one year ahead.

Dechow et al. (1995) and Kasznik (1999) found that firms with higher (lower) earnings exhibit significantly positive (negative) discretionary accruals, suggesting earnings management varies with earnings or that the Jones (1991) model used to estimate nondiscretionary accruals is misspecified.

Earnings Permanence and the Incremental Information Content of Cash Flows from Operations Cheng, Liu & Schaefer (1996), findings suggested that the incremental information content of accounting earnings decreases and the incremental information content of cash flows from operations increases, with a decrease in the permanence of earnings.

Lipe and Kormendi (1994) demonstrated that actual persistence measures derived from lower order ARIMA models did not capture all the value relevant time series properties of actual earnings because lower order models ignore a number of negative autocorrelations at higher lags that, though small individually were significant when considered in the aggregate.

Hopwood and McKeown (1992) studied the time series properties of quarterly operating CFs per share and earnings per share for a sample of manufacturing companies. This study finds the time-series properties of CFs are quite different from those of earnings. Results indicate a pattern of autocorrelation that is much stronger in the earnings series than in the CF series. In CF predictive ability tests, researcher compared firm specific ARIMA models to the premier ARIMA models attributed to Brown and Rozeff (1979) and Griffin (1977). The former models exhibited a slight advantage over the premiers. Unfortunately, these premier ARIMA models, which were originally identified on earnings data, may not have served as a useful benchmark because Hopwood and McKeown’s (1992) autocorrelation patterns for the CF series differed from the typical patterns for earnings series.

Freeman and Tse (1992) documented a nonlinear relation between abnormal returns and unexpected earnings. Researchers argue that as the absolute value of unexpected earnings increases, the "persistence" of earnings declines (Brooks and Buckmaster (1976) and Freeman, Ohlson, unexpected earnings from a linear model would predominantly reflect the effects of transitory, rather than permanent, earnings (because a linear model heavily weights the coefficient on high magnitude transitory earnings). Researchers shown that forcing a linear specification on an abnormal return un expected earnings model biases the slope coefficient on un expected earnings toward zero.

Livnat and Zarowin (1990) examined the cash flow components and report that the disaggregation of cash flow into its financing and operating components significantly improves the degree of association of cash flows with security returns.

Lorek & Willinger (1989) examined the differences in the auto regressive parameters of the Foster and Brown and Rozeff ARIMA models across firm-size strata. One-step-ahead quarterly earnings forecasts were generated by a set of best fitting time-series models. Research Tests also indicated that large and medium size firms generated one step ahead forecasts that were significantly more accurate than smaller firms at the .05 level and they obtained similar predictive findings on the significance of the size-effect in a supplementary analysis of the non seasonal and volatile growth and inconsistent strata membership firms.

Bernard and Stober (1989) found no evidence that the stock prices respond in a systematic manner to the release of information about the cash flow and accrual components of earnings and conjecture that the information content of these two components of earnings may not be systematically different.

Dharan (1987) examined the comparative abilities of accrual sales and cash collections of sales to predict future cash flows. It is found that when cash realization occurs in a period subsequent to sales realization, cash flow forecasts from earnings based on accrual sales are better than cash flow forecasts from earnings based on cash collections. This is because of accrual sales “provides information on management’s expectations about future cash flows (Dharan, 1987).

Financial analysts do not rely upon CF analysis; Analysts view it as an important supplementary tool useful in avoiding misleading inferences in the patterns of accrual based earnings numbers Dorfman (1987).

The magnitude of abnormal returns associated with good or bad news earnings signals is inversely related to firm size Freeman (1987), the researcher speculate that findings might simply be due to differential time series properties of the earnings numbers of large & small firms an uncontrolled factor in research design & calls for future research to examine the possibility. The empirical evidence on the importance of the size effect in the above settings, led to consider whether controlling explicitly for firm-size leads to inter-firm differences in predictive ability.

Theoretical and empirical work in accounting and finance has documented the importance of firm size when testing the information in security prices with respect to future earnings (Collins et al., 1987) and interested in assessing the information in security prices with respect to the predictive ability of earnings, their finding that price based earnings forecasts outperform time series forecasts by a greater margin for larger firms than smaller firms is of direct interest here. Their result implies that firm-size may help to explain inter-firm differences in the predictive ability of quarterly earnings data and helps to motivate the consideration of firm-size as an independent variable in the current study.

Freeman (1987) provided evidence that the magnitude of abnormal returns associated with good or bad news earnings signals is inversely related to firm-size. Freeman speculated that the findings might simply be due to differential time series properties of the earnings numbers of large and small firms an uncontrolled factor in research design and calls for future research to examine the possibility.

Brown et al. (1987) compared the accuracy of analysts to time series models based on historical earnings data. The attribute analysts’ superiority to a timing advantage (more information was publicly available if the forecasts were made after the public announcement dates) and an informational advantage (more information is used by the analysts than historical earnings data).

Barth, Cram & Nelson (1986) studied the role of accruals in predicting future cash flows. Findings proved that disaggregating earnings into cash flow and the major components of accruals significantly enhances earnings predictive ability, findings also showed relation between cash flow next year and current cash flow and each component of accruals is significant and has a sign consistent with prediction.

One of two researchers has re examined the association between earnings forecast error and earnings predictability because there was evidence suggesting that deliberate earnings forecast optimism is not an effective mechanism for gaining access to manager’s information ( Eames et al. 2002; Matsumoto 2002),For earnings level to be an important control variable in examinations of the association between forecast error and earnings predictability, there must be associations between earnings level and both forecast error and earnings predictability. Numerous studies report an inverse relation between forecast error and the level of reported earnings (Brown 2001; Eames et al. 2002; Eames and Glover 2002; Hwang et al. 1996). This association reflects earnings shocks due to unanticipated events and earnings management.

Olsen and Dietrich (1985) demonstrated that the monthly sales announcements of major department and discount stores provide information for investors not only for the retail giants but also for their suppliers. The sales volume announcements for the retailers furnish information on the future cash flow prospects for their suppliers and, thus, are incorporated into the suppliers’ share prices. Dharan (1987) investigated the comparative abilities of accrual sales and cash collections of sales to predict future cash flows. He shows that, when cash realization occurs in a period subsequent to sales realization, cash flow forecasts from earnings based on accrual sales are better than cash flow forecasts from earnings based on cash collections. This is because accrual sales “provide information on management’s expectations about future cash flows (Dharan, 1987, p. 445).”

Greenberg, Johnson, and Ramesh (1986) used 1963-82 compustate data to test the ability of earnings and CFFO to predict future CFFO, for each firm two separate ordinary least squares regression models were used. The first model test used previous earnings against current CFFO (earnings model) & the second model used CFFO for lags of 1 to5 years against current CFFO (cash flows model).R square for the earnings and cash flows model were compared and the model with the higher R square was determined to be the better predictor. The results showed that earnings outperformed CFFO in predicting future CFFO. It was concluded that the study provides evidence in support of the FASB's assertions that current earnings is a better predictor of future cash flows than is current cash flows.

Similarly a supporter of Greenberg et al's results is found by Murdoch and Krause (1990), However, the Singaporean study by Austin and Andrew (1989), whose approach was similar to that of Greenberg et al. (1986) found that neither earnings nor CFFO proved to be superior in predicting future CFFO.

Bowen, Burgstahler & Daley (1986) examined relationships between signals provided by accrual earnings and various measures of cash flow, Findings indicated that Correlations between traditional cash flow measures and alternative CF measures that incorporate more extensive adjustments are low, 2nd the correlations between alternative measures of CF and earnings are, while the correlations between traditional measures of CF and earnings are high. These first two results are consistent with earnings and alternative measures of CF that incorporate more extensive adjustments conveying different signals. Finally, for four out of five cash flow variables, the results are consistent with the hypothesis that random walk models predict CF as well as (and often better than) models based on other flow variables. An exception to this general result is that net income plus depreciation and amortization and working capital from operations appear to be the best predictors of cash flow from operations. Overall there results are not consistent with the FASB's statements that earnings numbers provide better forecasts of future cash flows than do cash flow numbers.

The firm size independently explains a substantial portion of the variation in post announcement drifts in security returns due to potentially misspecified quarterly earnings expectation models Foster et al (1984). (Ball and Watts 1972, Albrecht, Lookabill & McKeown 1977, Watts and Leftwich 1977 and Lev 1983 studied the Earnings ability to predict future earnings studied first or second order autocorrelations and or forecasts over one or two-year horizons and provided evidence to support a random walk model that is uncorrelated earnings changes, However, random walk may not be descriptive of the earnings process whereas Ramesh and Thiagarajan (1989) rejected a random walk earnings model and Lipe and Kormendi (1993) showed that higher order, rather than random walk, models are descriptive of market-adjusted earnings' time-series process. The magnitude of abnormal returns associated with good or bad news earnings signals is inversely related to firm size (Freeman 1987), speculates that these findings might simply be due to differential time-series properties of the earnings numbers of large and small firms-an uncontrolled factor in his research design-and calls for future research to examine the possibility.

Earlier additional information content of cash flows relies primarily on cross sectional regression models relating both earnings and cash flows to security return metrics that assumes a uniform relation between earnings (cash flow from operations) and security returns across observations. Ali (1994) however, conditions the incremental information content of unexpected earnings and cash flows from operations on their magnitude with respect to price. It is found that changes in earnings (cash flows from operations) are not expected to persist and thus have reduced implications for returns.

Dechow & Dichev (1986) suggested a new measure of one aspect of the quality of working capital accruals and earnings, Research illustrated the usefulness of analysis in two ways. First, they examined the relation between measure of accrual quality and firm characteristics. The accrual process suggests that the magnitude of estimation errors systematically related to business fundamentals as the length of the operating cycle and variability of operations. Researchers found that the accrual quality is negatively related to the absolute magnitude of accruals, the length of the operating cycle, loss incidence, and the standard deviation of sales, cash flows, accruals, and earnings, and positively related to firm size. Findings show that these observable firm characteristics can be used as instruments for accrual quality. The regression based accrual quality demands estimation of accrual quality demands long time series of data and the availability of subsequent cash flows that makes it costly or infeasible for certain practical applications (e.g. Quality of Accruals based trading strategies). Second Research illustrated the usefulness of analysis by exploring the relation between measure of accrual quality and earnings persistence. Firms with low accrual quality and more accruals those are unrelated to cash flow realizations so it produced more noise and less persistence in their earnings. Indeed they found a strong positive relation between accrual quality and earnings persistence. Although the measure of accrual quality is theoretically and empirically related to the absolute magnitude of accruals, and Sloan (1996) documents that the level of accruals is less persistent than cash flows. Probing further, they found out that accrual quality and level of accruals are incremental to each other in explaining earnings persistence, with accrual quality the more powerful determinant.

There are two widely held views regarding management’s motivations to managing earnings and each has quite different implications for the predictive usefulness of the resultant numbers .One view is that earnings management is motivated by mangers attempt to sustain the overvaluation of the firm’s stock price and to enhance managers personal welfare by disguising the true underlying economic performance of the firm (opportunistic perspective). Managers manage earnings to reveal relevant information about the future prospects of firms. They shown that originally reported (managed) earnings of firms classified as managing earnings for opportunistic reasons are less predictive of future cash flows relative to the restated (unmanaged) numbers. Conversely, they find that originally reported (managed) earnings of firms classified as managing earnings for informational reasons exhibit greater predictive ability with respect to future cash flows relative to restated (unmanaged) numbers (Collins and lys 2007).

Cheng & Dana (1996) examined the persistence of cash flow components in predicting future cash and the findings were that the cash flow components from various operating activities persist differentially. Findings were that the cash related to sales, cost of goods sold, operating expenses and interest persists a great deal into future cash flows; cash related to other has lower persistence and cash related to taxes has no persistence and then they incorporated accrual components into persistence regression model and found that the persistence of cash flow components are generally higher than those of accruals; however, accrual components do enhance model performance, their findings are consistent with the AICPA’s and financial analysts’ rationale for their recommendation that the financial effects of a company’s core and non-core cash flows should be distinguished.

Bowen et al. (1986) predicted CF from operations one and two years into the future by employing a set of alternative predictor variables including current net income, net income plus depreciation, working capital from operations and past values of CF from operations. Researcher confined data analyses to annual numbers and employed a limited set of simple linear forecasts. Researcher did not attempt to identify multivariate CF prediction models. The study found that the differences in relative forecast errors of the net income and CF from operations predictor variables are not significant none of this study results were consistent with the FASB's assertion of the superiority of earnings as predictors of future cash flows. However the use of relatively short, annual data bases in conjunction with naive expectation models limits the generalizability of this study.

Lev (1983) examined the cross sectional relation between a set of economic characteristics and the first and second order autocorrelation coefficients of earnings changes, return on equity changes, and sales changes. The main objective was to see if characteristics of the studied time series were related to the firm's economic environment. His finding that there were associations between characteristics of the firm's economic environment and the first two autocorrelation coefficients in earnings changes can be viewed as consistent with result that a persistence measure from a higher order (2,1,0) ARIMA model of earnings is associated with the characteristics of the firm's economic environment.

Brooks (1981) compared the predictive abilities of quarterly cash flows for a sample of 30 firms. The sample period was from 1964 to 1978. In this study univariate and transfer function Box- Function was used to procedure to develop forecasting models. In this article cash was defined for the quarter as earnings from operations fro the quarter plus depreciation and amortization in the quarter plus a quarter of the annual change in the deferred taxes. The input series in his multivariate model was earnings before extra-ordinary items; the study found the addition of earnings series to cash flow series in a multivariate setting did not improve the prediction of cash flows that were obtained from past cash flows series alone in a univariate setting. Thus there was no statistically significant difference between the two Box- Jenkins forecasting models second examining the residuals from the earnings univariate model and cash flow univariate model, on a firm basis, the residual mean square error was smaller for the earnings model than cash flow model, thus indicating that earnings model “ Fit” the earnings data better than cash flow model to cash flow data.

Foster (1981) examined the impact of earnings announcements on the security prices of other firms in the same industry. Foster documented statistically significant security returns for non announcing firms in ten industries at the time of "large" security returns for announcing firms within the same industry. These "intra industry information transfers" suggest news releases of other firms within an industry are used in the determination of a given firm's security price.

Khumawalla (1978) also used the Box-Jenkins univariate procedure to provide the time series properties of quarterly Cash Flows for a sample of 29 airline companies. In this study researcher compared the aggregation of the data on the predictive ability of quarterly cash flows. Sample consisted of thirty airlines that was covering the period 1965-76. The Cash flow was defined as cash flow from operation, but she did not included minority interest in the calculation of quarterly cash flow and the whole of subsidiary income recognized under the equity method of accounting was treated as unremitted earnings.

Albretch, Lookabill and Mckeown (1977) examined the time-Series properties of undeflated earnings and earnings deflated by shareholders equity. The studied utilized only twenty – five observations for the 1947-75 period to avoid the problems of structural changes. Study sample consisted of 49 firms from three industries and researcher also compared the one, two, and three years ahead predictions from the firm specific models with a random walk with a drift model. Utilizing five error measures, mean relative error, mean absolute relative error, mean squared relative error and average ranking, researcher found that random walk with a drift performed as well as the firm specific models for the undeflated earnings and random walk outperformed the firm specific models fro the deflated earnings.

Foster (1977) examined the behavior of quarterly earning, sales and expenses series of 69 forms over the 1946-74 period on a cross sectional basis, in this study the author examined the predictive ability of six forecasting models to forecast one period ahead for each quarter from 1962-74. The models included two simple seasonal quarter by quarter models, two simple adjacent quarter models, a model suggested by him , and firm specific models identified by using the Box- Jenkins techniques. Researcher suggested model had seasonal fluctuations and a trend in the time series and utilized three error metrics: average rank, mean absolute percentage error, and mean squared percentage error then researcher applied the Friedman analysis of variance test where the rank of one was assigned to the most accurate forecasts in any given period. In the end Foster concluded that the model had the lowest rank in each quarter. Moreover, Box- Jenkins models were outperformed by his model when considering the sales and expenses series.

Griffin (1977) also examined quarterly earnings, sample consisted of 94 firms. In this study four models representing a broad range of linear auto regressive integrated moving average ARIMA models. Study presented some preliminary evidence and its implications for accounting research and security prices. By applying cross-sectional analysis researcher concluded that “Quarterly earnings may be parsimoniously. One reflects the adjacent quarter movement and the other reflects the quarter to quarter movement over time” .


Is there a significant relationship between Share Price with the three independent variables (i.e. Cash Flow, Sales, and Earning)?


What is the relationship of Earning, Cash Flow & Sales with Share Price?


In the following section hypothesis are generated which describes Section 4.1, Section 4.2 describes data collection and section 4.3 describes the research method used in this study.

4.1 Hypothesis Development

H1: There is an association between Cash Flow from Operation & Share Price.

H2: There is an association between Sales & Share Price.

H3: There is an association between Earning & Share Price.

4.2 Data Collection

To evaluate the relationship of Cash Flow from operations, earnings and current operating cash flows with share price, sample of 16 companies of Cement sector which are Listed at KSE (Karachi Stock Exchange), data obtained from Karachi Stock Exchange, SBP (State bank of Pakistan), Financial statements of companies and cement company website and collected the annual Share price value of each company from the Karachi Stock Exchange and company’s website. The data collected included earnings (as measured by after-tax income from operations), operating cash flows, and net sales for the year of 2003, 2004, 2005, 2006, 2007 and 2008.

For example, financial year ends in June and share price values were taken as annual, Similarly Cash Flow, Sales and Earning. The variables were examined in SPSS in Pak Rupee.

4.3 Research Method

To evaluate the relationship of variables the explanatory variables were regressed using Ordinary Least Square Regression Analysis (OLS). Share price was chosen as the dependent variable not only because of its obvious importance in the financial literature but also because it proxies for a firm’s future cash flows in the broadest sense.


This section consists on 3 Tables, table 5.2 presents Model Summary, Tables 5.3 Coefficients and Table 5.4 Anova.

5.1 Regression Equation

Share Price=α+β1 (Cash Flow from Operation) + β 2 (Sales) + β 3 (Earning) +µ


Model Summary



R Square

Adjusted R Square

Std. Error of the Estimate






A. Predictors: (Constant), Profit After Tax, Sales, Cash Flow from Operation

The model summary shows us that coefficient of correlation is 0.712 which means there is a positive relationship of Cash Flow, Sales and Earning on Share Price. Coefficient of determination is 0.507 or 50.7% which means the regression model explains 50.7%.


Coefficients (a)


Unstandardized Coefficients

Standardized Coefficients




Std. Error








Cash Flow From Operation












Profit After Tax






A. Dependent Variable: Share Price

Share Price = 20.480+ -4.12 (Cash Flow from Operation) + 1.33(Sales) +3.60(Profit after Tax).

When all independent variables are zero than the value of the Share Price will be 20.480.

If the value of Cash Flow from Operation changes by 1% than the value of Share Price will Decrease by 4.12% and significance value of Cash Flow from operation is .061 it means hypothesis is accepted and relationship does not exists between Cash Flow from Operation and Share Price.

If the value of Sales changes by 1% than the value of Share Price will Increase by 1.33% and significance value of Sales is 0.108 it means hypothesis is accepted and relationship does not exists between Sales and Share Price.

If the value of Earning changes by 1% than the value of Share Price will increase by 3.60% and significance value of Earning is .000 it means hypothesis is rejected and relationship exists between Earning and Share Price.




Sum of Squares


Mean Square

















A. Predictors: (Constant), Profit After Tax, Sales, Cash Flow from Operation

B. Dependent Variable: Share Price

Model is explained by 41152.641 (50.7%) and remaining 40075.067(49.3%) is residuals. Significance value is .000 its means model is moderately fit.


Previously studies did not focus on sales as predict variable of share price because entity’s of operating cash flows derives eventually from its sales. The accrual components in the previously studies included the change in accounts payable, change in inventory, depreciation and others. This study empirically test the relationship of cash flows, sales , earning with share price and whether earnings , sales or cash flows measure are a better predictor of share price though earning is a determinant of share price. The results support the following conclusions.

Operating cash flows from operation does not appear to be significant and better predictor of share price because its relationship with share price seemed insignificant, though it was also seen inconsistent in previous study.

In terms of prediction and relationship sales also appears insignificant with Share Price, in current study sales considered as one of the accrual components.

Earning seemed significant and it has positive relationship with share price and its relationship with share price is significant and it predicts better than cash flows & Sales In previously studies earning has been better predictor of share price and has been positive correlated with share price.

Need help with your literature review?

Our qualified researchers are here to help. Click on the button below to find out more:

Literature Review Service

Related Content

In addition to the example literature review above we also have a range of free study materials to help you with your own dissertation: