Close

Sample financial essay

  1. Introduction

The finance theory asserts that the value of the equity shares of a company depends upon the future free cash flows that are attributable to the shareholders of the company. Every analyst and investor attempts to forecast the future cash flows and discount them at appropriate discounting rates in order to arrive at the fair value of equity shares which are then compared to the current market price in order to arrive at investment decisions based on the extent to which the shares are under valued or over valued as per the current market price. The primary set of information that are used for valuation are the earnings performance of the company as presented in the annual and the quarterly reports filed by the company. Quite interestingly while the valuation takes place on the basis of cash flows, the accounting systems work on the basis of accruals. The biggest advantage of using cash flows for valuation is that it may be possible for the management to artificially inflate the earnings using accruals but it is extremely difficult to influence cash flows artificially. However over the long term it is essential for the earnings and the cash flows to follow a largely aligned path even though in short term there may be some deviations. The  purpose of this brief research paper is to critically examine the usefulness of cash flows and the accruals in predicting future earnings of the companies. This report is split into five parts. The second part reviews some of the important research papers that have focused on this topic in the past. The third section is focussed on presenting the methodology that would be adopted by this research to arrive at the conclusions. The fourth section presents the findings drawn on the basis of the empirical analysis performed in this research. The final section draws conclusions on the basis of the findings.

 

  1. Literature Review

The purpose of this research is to critically examine the relationship that future earnings has with accruals and cash flows. Several researchers have attempted to study this topic in the past and have drawn some important observations. The purpose of this section of the report is to perform a review of some of the important research papers in this topic of interest and to arrive at observations on the basis of the conclusions drawn by these researchers. Even though this research uses a sample drawn from the US market for performing the empirical analysis, the review of literature is performed on a set of papers that may have studied different markets including the US market. This helps to ascertain if there are significant variations in the conclusions drawn by researchers on the basis of the sample of markets includes in their research.

 

Dechow (1994) studies the impact of accruals in influencing the ability of earnings to measure firm performance. The firm performance is indicated by the returns generated by the stock of the companies. The author notes that there are some important timing and matching differences between accruals and cash flows because of which the ability of earnings to predict future stock performance is impacted. The author concludes that the relevance of accruals is increased when the performance measurement interval is short, the operating cycle of the company is longer and the volatility in working capital requirement is higher. The conclusions drawn by the researcher on the basis of the empirical analysis is well in line with the theoretical factors that could cause an increase in accruals. It is seen that when there is an increase in volatility of working capital or an increase in operating cycle of the company there could be a widening difference between the earnings and the cash flows of the company as accruals tend to increase due to these reasons. Farshadfar and Monem (2013) re-examine the hypothesis that the operating cash flows could help to improve the predictive ability of the earnings reported by the companies. The authors use the empirical data drawn from the Australian sample of listed companies. Using the results of empirical analysis the authors conclude that the components of operating cash flows have more predictive ability than earnings or total accruals. These results are found to be robust to variations in industry membership, profitability of the firm and size of the firm. Overall this is an important research paper which asserts that there is more predictive value in the component of operating cash flow than the headline numbers that are typically reported widely by the companies.

 

Francis et al. (2005) note that there is significant level of information risk involved in the annual accounting data shared by the companies with their shareholders. The authors also note that the accruals used in accounting reporting form the primary source of such risk. Therefore the authors attempt to ascertain if the market price of the shares includes a pricing done for the quality of accruals reported by the management. According to the empirical analysis and the results obtained thereof the authors note that the market tends to price the quality of accruals in a significant manner. Specifically the poorer accrual quality is directly linked to higher cost of equity for the companies. It is also observed that the companies with poor quality of accruals tend to see an increase in cost of debt as well. These conclusions are found to be consistent even when different measures of accrual quality was used by the researchers. Then the authors split the accruals into discretionary accruals (which could be controlled by the management) and non discretionary accruals (which are driven by the market factors). The quality of non discretionary accruals seem more important than that of the discretionary accruals as per the empirical analysis’ results.

 

Leuz et al. (2003) study the relationship between corporate governance and the quality of earnings reported by the management using samples drawn from 31 countries. According to the authors the management and the insiders tend to use earnings management practices as means to create information asymmetry which in turn results in poor corporate governance as it is widely recognised that a wider distribution of ownership of the companies’ equity shares is associated with superior corporate governance. When there is a high level of earnings management the shares tend to be owned by a small group of insiders and hence the authors assets that there is an inverse relationship between distribution of ownership and the quality of corporate governance in the companies. Lundholm and Myers (2002) study the relationship between the current earnings reported by the management and the future earnings of the firm. According to the authors the level of disclosures made in the annual reports impact the extent to which this relationship could be ascertained by the shareholders. Typically when the disclosure levels are high there is a strong and predictable relationship between the reported earnings and the future earnings. On the other hand the companies that do not disclose sufficient level of information tend to have one a weak relationship between the current and future earnings. Thus according to the researchers, the disclosure levels affect the predictability of the future earnings.

 

Several researchers have studied and concluded that the details within the accruals could lead to better levels of prediction than the aggregate earnings numbers. Barth et al. (2001) study the ability of accruals and the various components to predict future performance of the companies. Based on the empirical analysis the authors conclude that the components within the accruals have better ability to product future cash flows than the earnings numbers. The authors argue that the aggregate earnings numbers tend to mask several pieces of information which in turn reduce the level of predictability and usefulness. On the basis of these the authors argue that the components within non discretionary accruals tend to be more useful than earnings in helping the analysts predict future earnings.

 

From the review of literature it is noted that most of the researchers who have studied the predictive value of earnings and cash flows have concluded that the accruals have better ability to predict future earnings than the current period earnings. Further they also note that the details within the accruals could help in predicting future performance better than the headline numbers. The next chapter of this report provides details on the methodology that is adopted to study the empirical evidence in relation to the predictive ability of earnings and cash flows.

 

  1. Methodology

The purpose of this chapter of the report is to present details on the methodology that is adopted by this research to meet the objectives. It is noted from the review of literature that most of the researchers who have studied this topic have used empirical methods to arrive at their findings and conclusions. The biggest advantage of empirical research method is that it allows the researcher to use the actual data drawn from the market to test the theoretical models. The use of past data helps to validate the relevance of the theoretical models. Empirical results tend to have more practical use than the other methods because of the fact that the actual past data is used for validation. The empirical research method requires an appropriate model to be tested using the empirical data. In this case the research is focused on testing the predictive ability of the earnings and the accruals. Therefore the following model is used for testing.

NIit+1 = ai + b1 CFOit + b2 TotalAccrualit + b3 TAit + ei

Where

NI is net income reported by the companies in their annual reports

CFO is Cash from Operations reported by the companies as part of their cash flow statements

TotalAccrual is the total accruals implied in the financial statements. Total accruals is estimated as the difference between Net income and cash from operations.

TA is total assets, which is a control variable.

t is the time period in years

As it can be seen that this model has both cross sectional and time series data to be estimated it is important to use a panel data regression method to estimate this model.

 

In the above model it is possible that there is a significant level of correlation among the NI, TA and Total Accruals variables as bigger companies tend to have higher values for all these three variables and smaller companies are likely to have smaller values for these variables. The following model is also estimated alongside. In this model the net income, CFO and total accrual variables are divided by total assets in order to adjust for the variations in sizes of the companies.

 

NI/TAit+1 = ai + b1 CFO/TAit + b2 TotalAccrual/TAit + ei

 

Since NI includes both CFO and total accruals it may be interesting to see which component of NI could predict the future performance better.

 

This analysis used empirical data drawn from the US market. The components of both S&P 500 and Nasdaq 100 are included in the sample. Since there is some overlap between these two indices the total numbers of stocks included in the sample is 527. The sample covered the annual financial data reported by these companies between 2008 and 2017, a period of 10 years. The data for this research is drawn from Compustat through WRDS. The data is expected to be free from errors and omissions as the database is professionally managed. Stata is the statistical analysis software used for performing all the analysis as part of this research.  It is noted that Stata has strong capabilities to perform panel data regression analysis on large datasets.

The next chapter of the report is concerned with presenting the results of the empirical analysis and the observations drawn on the basis of these findings.

 

  1. Findings and Analysis

The purpose of this chapter is to present the results of the regression analysis performed on the empirical data. This chapter is divided into two sections. The first section presents the descriptive statistical summary of the empirical data and the second section presents the results of the estimation made using panel data regression method.

 

Following table shows the descriptive statistical properties of the sample used for analysis.

 

Variable |        Obs        Mean    Std. Dev.       Min        Max

————-+———————————————————

TA |      3,608    55828.34      189063     51.699    2573126

NI |      3,607    1585.072    4252.453     -30860     104821

CFO |      3,606    3020.127    7009.388     -55705     121897

TotalAccrual |      3,750   -1379.525    5785.828    -123827     122153

 

It can be seen that the total observations vary for each variable as there are some unreported values for some companies for some years. All these data are removed from the sample for the purpose of regression analysis. It is seen that the average total assets for all the firms combined is about USD 55.83 billion while the standard deviation is quite high at USD 189 billion. This shows that the sample includes quite a large variation of sizes of companies even though all these companies are part of the major indices in the US. The largest company in the sample had a total asset base of USD 2.5 trillion and the total assets of the smallest company in the sample was USD 52 million. The average of net income reported by these companies was USD 1.585 billion with a standard deviation of USD 4.25 billion. The sample included both loss-making and profit-making companies. The cash from operations (CFO) averaged about USD 3.02 billion with a very high standard deviation of USD 7.01 billion. Total accrual which is estimated as the difference between net income and CFO averaged about USD -1.4 billion for the overall sample. The standard deviation in total accruals was USD 5.79 billion. It is important to note that the reported net income is less than CFO for most of the cases making the total accruals negative.

Following table shows the results of the random-effect panel data regression performed on the following model, as explained in the third chapter of this report.

NIit+1 = ai + b1 CFOit + b2 TotalAccrualit + b3 TAit + ei

 

Random-effects GLS regression                   Number of obs     =      3,200

Group variable: gvkey1                          Number of groups  =        405

 

R-sq:                                           Obs per group:

within  = 0.0089                                         min =          2

between = 0.8545                                         avg =        7.9

overall = 0.3746                                         max =          9

 

Wald chi2(3)      =    1914.43

corr(u_i, X)   = 0 (assumed)                    Prob > chi2       =     0.0000

 

——————————————————————————

F.NI |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]

————-+—————————————————————-

CFO |   .4722042   .0152273    31.01   0.000     .4423593    .5020491

TotalAccrual |   .2788669   .0160278    17.40   0.000      .247453    .3102808

TA |   .0025313   .0004023     6.29   0.000     .0017429    .0033197

_cons |   527.9694   64.86312     8.14   0.000       400.84    655.0987

————-+—————————————————————-

sigma_u |          0

sigma_e |  2870.4551

rho |          0   (fraction of variance due to u_i)

——————————————————————————

 

It can be seen from the table above that the coefficient of CFO is 0.4722 with a p-value of 0.000. The null hypothesis that is tested in this case is that the coefficient is not different from zero. Since the p-value is less than 0.05 the null hypothesis is rejected at 5% significance level. This shows that the CFO has a significant positive coefficient. The total accrual variable has a coefficient of 0.2788 with a p-value of 0.000. Here too it is noted that the p-value is less than 0.05 and hence the coefficient is significant. It is interesting to note that CFO has a higher significant value than total accruals which shows that the CFO predicts the future earnings better than the accruals do. The total assets variable has a coefficient of 0.0025 which is also found to be statistically significant at 5% level.

 

The Chi-square statistic is estimated to be 1914.43 with a p-value of 0.000. The null hypothesis that is tested in this case is that all the independent variables included in the model have coefficient values of zero. However since the null hypothesis is rejected it could be concluded that at least one of the independent variables has significant coefficient as per the Chi-square value. The r-squared value measures the extent to which all the independent variables together explain the variations in the dependent variable. In this case it is seen that the overall r-squared value is 0.3746 which means that about 38% of the variations in the future earnings could be explained by the three independent variables together.

 

Following table shows the results of the panel data regression performed on the second model shown below.

NI/TAit+1 = ai + b1 CFO/TAit + b2 TotalAccrual/TAit + ei

This model adjusts for the size impact by dividing all the variables by total assets.

 

 

Random-effects GLS regression                   Number of obs     =      3,200

Group variable: gvkey1                          Number of groups  =        405

 

R-sq:                                           Obs per group:

within  = 0.0507                                         min =          2

between = 0.7988                                         avg =        7.9

overall = 0.3827                                         max =          9

 

Wald chi2(2)      =    1982.39

corr(u_i, X)   = 0 (assumed)                    Prob > chi2       =     0.0000

 

——————————————————————————–

F.NI_1 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]

—————+—————————————————————-

CFO_1 |   .6436254   .0147386    43.67   0.000     .6147383    .6725125

TotalAccrual_1 |   .3248462   .0152899    21.25   0.000     .2948785    .3548139

_cons |   .0068563   .0018604     3.69   0.000     .0032099    .0105026

—————+—————————————————————-

sigma_u |          0

sigma_e |  .05447495

rho |          0   (fraction of variance due to u_i)

——————————————————————————–

 

From the above table it can be noted that the total observations used for analysis is 3200 and this covered the data relating to about 405 companies. On an average each company had about 8 years of data and the minimum was 2 years while the maximum was 9 years. It is seen that the coefficient of CFO/TA is 0.6436 with a p-value of 0.000. The p-value clearly shows that the coefficient is significant at 5% level. The positive coefficient also means that the increase in CFO/TA tends to lead to increases in next year’s net income/TA. It is observed that the coefficient of Total accruals/TA is 0.3248 with a p-value of 0.000. Thus this coefficient is also found to be statistically significant at 5% level. It is interesting to note that the Z-statistic of Total accrual/TA is less than that of CFO/TA which means that the latter has a stronger influence on the dependent variable than the former. Thus on the basis of these results it is observed that the accruals have less predictive power than CFO. The overall r-square value is observed to be 0.3827 which means that the independent variables could together explain about 38% of the variations in the future earnings. Even though this seem quite low in comparison to the theoretical limit of 100%, it is common for the empirical models to have low r-squared values as there may be various other factors that could influence the dependent variable. The chi-squared statistic is 1982.39 with a p-value of 0.000. Since the p-value is less than 0.05 the null hypothesis that none of the coefficients are significant could be rejected at 5% level.

 

Following table presents the results of the same regression performed with two-year lead Net Income as the dependent variable. In other words the following model tests the ability of the independent  variables to predict the second year net income/TA.

 

Random-effects GLS regression                   Number of obs     =      2,795

Group variable: gvkey1                          Number of groups  =        405

 

R-sq:                                           Obs per group:

within  = 0.0137                                         min =          1

between = 0.5976                                         avg =        6.9

overall = 0.3056                                         max =          8

 

Wald chi2(2)      =     621.21

corr(u_i, X)   = 0 (assumed)                    Prob > chi2       =     0.0000

 

——————————————————————————–

F2.NI_1 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]

—————+—————————————————————-

CFO_1 |    .450264   .0180698    24.92   0.000     .4148478    .4856801

TotalAccrual_1 |   .1071025   .0169786     6.31   0.000      .073825    .1403799

_cons |   .0186567   .0024173     7.72   0.000      .013919    .0233944

—————+—————————————————————-

sigma_u |  .02092881

sigma_e |  .04983699

rho |  .14991577   (fraction of variance due to u_i)

——————————————————————————–

 

It is seen from the above results that total accruals/TA and CFO/TA could explain even the second year earnings with significant effect. However the overall strength of the model itself is seen to be less than the first year prediction model as the r-squared value is less in this case.

 

It is seen from the review of literature that some researchers including Dechow (1994) have argued that accruals could have some significant predictive powers with respect to the future earnings. On the other hand some other researchers including Farshadfar and Monem (2013) have argued that operating cash flows could help to predict the future earnings more accurately than the accruals or the current period earnings. Based on the results of empirical analysis performed in this chapter it is evident that the CFO has better predictive ability than total accruals even though both could predict the future earnings at significant level. It is also noted that these variables could predict even the earnings two period down the line.

 

  1. Conclusion

The purpose of this final chapter of the report is to present a brief summary of the research and also draw important conclusions on the basis of the findings presented in the previous chapter.

 

It is noted that several researchers in the past have focussed on the predictive ability of the accruals and the cash flow. The purpose of this research is to critically examine and compare the ability of cash flow from operations and the total accruals to predict the future financial performance of the companies. From the review of literature it is noted that most of the researchers have found the accruals to have more predictive ability than the CFO while some have found CFO to be more useful in predicting future period performance of the companies. This research uses empirical research method and tests the model using data drawn from the US market. The data sample used includes a set of 540 companies that form part of the S&P 500 index or Nasdaq 100 index.

Based on the results obtained it can be concluded that both total accruals and CFO have significant ability to predict future net income of the companies. These results are found to be consistent even after adjusting for the size effect of the companies. It is important to note that the predictive power of the CFO is much higher than that of total accruals. Even though some researchers have concluded that accruals have better predictive abilities these conclusions seem to be more in line with conclusions made by a smaller set of researchers including Farshadfar and Monem (2013).  The most important implication of this research is that the analysts and the investors in the market are better off focussing on CFO as one of the important factors for investment decision-making or forecasting of next period’s financial performance of the companies. More interestingly the research shows that the CFO and accruals have ability to predict even the second next period earnings as well.

 

There are some important ways in which this research could be further improved. It is noted that this research has used data pertaining to about 9 years. It would be interesting to repeat the research for each year separately to note if there are any significant year effects that could be found in the results. Further it is well known that the accruals tend to follow a rise-fall pattern as the gap between CFO and net income widens and narrows periodically. Further it may also be interesting to repeat this research with different sets of samples drawn from different markets – both developed and developing in order to ascertain if there are any significant market effects that are embedded in these results. It may also be helpful to include some more variables in the model such as the industry classification, market leadership position of the companies etc., as these factors could have some significant impact on the overall earnings of the companies.

 


References

 

 

Barth, M.E., Cram, D.P. and Nelson, K.K., 2001. Accruals and the prediction of future cash flows. The accounting review, 76(1), pp.27-58.

 

Dechow, P.M., 1994. Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals. Journal of accounting and economics, 18(1), pp.3-42.

 

Farshadfar, S. and Monem, R., 2013. The usefulness of operating cash flow and accrual components in improving the predictive ability of earnings: a re‐examination and extension. Accounting & Finance, 53(4), pp.1061-1082.

 

Francis, J., LaFond, R., Olsson, P. and Schipper, K., 2005. The market pricing of accruals quality. Journal of accounting and economics, 39(2), pp.295-327.

 

Leuz, C., Nanda, D. and Wysocki, P.D., 2003. Earnings management and investor protection: an international comparison. Journal of financial economics, 69(3), pp.505-527.

 

Lundholm, R. and Myers, L.A., 2002. Bringing the future forward: the effect of disclosure on the returns‐earnings relation. Journal of Accounting Research, 40(3), pp.809-839.

2017-08-29T15:58:36+00:00