«Leif Atle Beisland University of Agder Dissertation submitted to the Department of Accounting, Auditing and Law at the Norwegian School of Economics ...»
Note that multicollinearity could have been a challenge in these specifications since four related measures are used as explanatory variables. Fortunately, it is not. The variance inflation factor (VIF) is computed for each regression, and they are each significantly below the cut-off threshold of 10 proposed by for instance Hair et al. (2006). As a result, multicollinearity is not considered a problem in the regression analyses.
Overall, the findings of step 1 indicate that both cash flow and accruals are significantly related to future firm performance. However, the results are dependent on the sign of earnings and whether cash flow or earnings is forecasted. The most unambiguous results are found for cash flow levels. Present cash flow is related to future cash flow and earnings in the total sample and when earnings are positive. It does not matter whether it is next year’s performance or the mean of the next three years that is analysed. Accruals are related to future earnings as long as earnings are not negative. However, accruals are generally not related to future cash flow. Except for short term predictions in the positive earnings sample, the change variables are insignificant. In the negative earnings sample, all explanatory variables are typically insignificant. Thus, cash flow and accruals seem to show little association with future firm performance when earnings are negative.
cash flow and earnings as the dependent variable of the regression specification. The results from this regression are presented in table 3. In this specification, all explanatory variables come up with significant coefficients in the total sample. Both cash flow and the change in cash flow are positively related to stock return. A high and increasing cash flow is generally associated with high stock return. This result is comparable to Easton and Harris (1991), who found that both earnings and the change in earnings generally are significantly related to stock return. Also analogous to Easton and Harris’ results on earnings, I find that the cash flow coefficient is higher than the change in cash flow coefficient. The coefficients on the two accruals variables are also significant. Since accruals typically are a negative earnings item, both the level and change of accruals appear to be negatively related to stock return. In other words, the findings suggest that large and increasing accruals are associated with lower stock return. In particular, increasing accruals seem to be a perceived as a negative signal by equity investors (see the large coefficient on this variable). To illustrate the significance of accruals in this specification, it is worth mentioning that the explanatory power is halved if accruals are deleted from the specification (not tabulated). Note that the explanatory power as measured by the adjusted R 2 is generally much lower in this analysis than in the first step of the study. Most variation in stock returns is explained by factors other than accounting variables such as earnings. This is a common conclusion in international value relevance research (compare Basu, 1997; Collins, Kothari, Shanken, & Sloan, 1994; Kormendi & Lipe, 1987; Baruch Lev, 1989).
Table description Table 3 describes the value relevance of earnings split into cash flow and accruals for a sample of Norwegian firms in the period 1992 to 2004. It summarises the regression coefficients (Coeff.), White-adjusted t-values (tvalue), total explanatory power (adj. R2) and number of observations (n) for the total sample as well as for the
positive and negative earnings sub-samples. Data are analysed using the following regression specification:
(2) RETi,t = β 0 + β1CFi,t + β 2 ∆CFi,t + β 3 ACC i,t + β 4 ∆ACC i,t + ε i,t where RETi,t is the stock return for company i in year t, CF is cash flow from operations and ACC is total accruals. ∆ denotes yearly change in the variables. The accounting variables are scaled by the market value of equity at 30 December in year t-1. Coefficients marked in boldface denote a statistical significance at a 5 % level, two sided test.
All other things being equal, my findings suggest that for given earnings, investors prefer higher cash flow proportions. More cash flow and less accruals are “better” as measured by stock returns. Sloan (1996) found that investors fixate on earnings and fail to fully reflect the information contained in the accrual and cash flow components of earnings (the accrual anomaly; see Baruch Lev & Nissim, 2006). Contrary to Sloan (1996), my findings suggest that investors understand the different information content and consider this when valuing equity. Sloan reports that investors price the cash flow component and the accrual component of earnings identically, even if the cash flow component is far more persistent. The higher persistence of cash flow than accruals is also found in my sample. However, my findings suggest that investors see through this phenomenon and price the two earnings components differently. The regression coefficients for cash flow and accruals are significantly different from each other on a 0.01 % level (untabulated).
highly associated with stock return. All regression coefficients are significant when earnings are positive. Explanatory power equals 13 %. The adjusted R 2 drops to 2 % when earnings are negative. Only the change in accruals shows a significant regression coefficient. Briefly summarised, stock returns seem to be associated with both cash flow and accruals. However, further analysis shows that the association typically only applies when earnings are positive.
5.3 Step 3: The Association Between Predictive Ability and Value Relevance The prediction from section 3 says that if accruals and cash flows are related to short-term future firm performance as measured by accounting earnings and cash flow, it is reasonable to expect that they are also significantly associated with current stock return. Current cash flow is significantly related to both future cash flow and future earnings, while current accruals appear to be significantly related only to future earnings. Both variables are value relevant. As such, one may claim the empirical findings support the proposed prediction. However, the regression coefficients are more significant in the value relevance study than in the future performance study. This means that cash flow and accruals have a closer association with stock return than with future cash flows or earnings (as measured by significance level of individual coefficients). Even if all explanatory variables seem significantly related to stock return in the total sample, none of the regressions in step 1 find that all variables are significant predictors of cash flow or earnings. Only when earnings are positive and one-year ahead cash flow or earnings are predicted do all explanatory variables become significant.
Taken together, steps 1 and 2 reveal that the prediction of section 3 cannot be reversed. Cash flow or earnings prediction studies may provide indications with respect to value relevance, but they do not at all present the complete picture. What I have referred to as indirect value relevance studies do not seem to be equivalent to “pure” or direct value relevance studies.
based upon short term predictions tests (indirect value relevance studies). Assume, for instance, that only cash flow predictions were studied, and that these studies should proxy for value relevance studies. Such tests would have suggested that accruals are not related to company value. The value relevance study shows that this conclusion would have been wrong.
An empirical result of particular interest is the finding that accruals are related to stock return even if they are not associated with future cash flow in the short term. From a “cash is king” perspective, one would have expected that the failure of accruals to predict future cash flow would render them unrelated to current stock return; see the discussion of Kim and Kross (2005) on the relationship between cash flow predictions and value relevance. However, even if accruals are unrelated to future cash flows, they appear to be related to future earnings.
Section 3 showed that company value could be expressed as a function of accounting earnings; compare the residual income model (Edwards & Bell, 1961; Feltham & Ohlson, 1995; Ohlson, 1995). As such, there might be a relation between accruals’ role as an earnings predictor and their value relevance. In addition, earnings have an indirect role in equity valuation, as many investors predict earnings and derive future cash flows from the earnings predictions. This issue is discussed by Lev et al. (2005). They state that “…while economic theory prescribes that asset values are determined by their future cash flows, financial analysts predominantly predict earnings” (B. Lev et al., 2005, p. 9). And further: “The underlying heuristics are somewhat obscured; perhaps investors predict earnings first, and derive future cash flow estimates from the predicted earnings” (B. Lev et al., 2005, p. 9). Lev et al. ask why investors and analysts have this almost universal “obsession” with earnings. They present one answer when showing that the returns on portfolios constructed from a perfect prediction of
cash flows. However, this result holds only for perfect predictions. When portfolios are constructed from actual and not perfect predictions, the returns are higher for cash flow than for earnings-based portfolios. Lev et al. suggest that some analysts may ignore the latter finding.
Kim and Kross (2005) are surprised to find that the over-time development in earnings’ ability to predict one-year ahead cash flow is not identical to the over time development in earnings’ value relevance. In fact, while accounting earnings seem to have lost value relevance over time (S. Brown, Kin, & Lys, 1999; Collins et al., 1997; Gu, 2007; Baruch Lev & Zarowin, 1999), Kim and Kross document that earnings’ ability to forecast cash flows has actually increased. To further investigate the possible relationship between short term cash flow and earnings predictions and value relevance, I examine if this phenomenon is present in my data sample as well. However, while Kim and Kross only focus on one-year cash flow predictions, I continue to analyse both three-year future cash flows as well as one- and threeyear future earnings. I split the sample in two, using 1999 as the cut-off year. This year is not randomly chosen as the cut-off year: The Norwegian Accounting Act of 1998 was put into effect in 1999. However, the Accounting Act of 1998 did not introduce any revolutionary changes in the Norwegian accounting system. The main principle is still historic cost with traditional principles for revenues and cost recognition, such that revenues should be earned and costs matched with the earned revenues for the period (Gjerde et al., 2007b). The most notable change was probably that fair value for liquid short term financial instruments was introduced. The partial effect of such a change should normally be increased value relevance of the accounting figures. Table 4 lists the results for the periods before and after 1999.
- 115 Table description Table 4 describes the predictive ability and value relevance of earnings split into cash flow and accruals for a sample of Norwegian firms in the period 1992 to 2004. It summarises the regression coefficients (Coeff.), White-adjusted t-values (t-value), total explanatory power (adj. R2) and number of observations (n) for the total sample as well as for the positive and negative earnings sub-samples. Panel A shows the results for the period 1992 to 1998. Panel B shows the results for the period 1999 to 2004.
Predictive ability (step 1) is analysed using regression specification (1a) to (1d), while value relevance (step 2) is analysed using regression specification (2):
where CFi,t is cash flow from operations for company i in year t, ACC is total accruals, EARN is earnings before extraordinary items and RET is stock return. ∆ denotes yearly change in the variables. The accounting variables are scaled by the market value of equity at 30 December in year t-1. Coefficients marked in boldface denote a statistical significance at a 5 % level, two sided test.
samples. I am focusing primarily on the results for the total sample. Table 4 reveals that the adjusted R 2 is higher for all cash flow and earnings predictions (step 1) in the first period than in the second period. It appears that current cash flow and accruals were a far better indicator of future firm performance as measured by cash flow and earnings in the 1992-98 period than in the 1999-04 period. Even so, current cash flow and accruals have a higher association with current stock return in the second period than in the first (step 2). Even if cash flow and accruals’ predictive ability for future firm performance have decreased over time, value relevance has actually increased26. This is further evidence that short term cash flow and earnings prediction tests are not equivalent to value relevance analysis. The analysis of predictive ability is more extensive here than in Kim and Kross’ (2005) study, as longer term cash flows as well as future earnings are added to the analysis. Still, the findings are consistent with Kim and Kross’ conclusions. The over time development in earnings’ predictive ability is not identical to the over time development in earnings’ value relevance.
Note, however, that my results are opposite in sign to those of Kim and Kross since the predictive ability has decreased while value relevance has increased in my sample27.
Prior research has shown that adjusted R 2 may be incomparable across samples (S. Brown et al., 1999; Gu, 2007). Specifically, Brown et al. (1999) and Gu (2007) show that scale differences and/or sampling variations might lead to adjusted R 2 differences even if the underlying economic relation is identical in two samples. The analysis of this section is The result that value relevance is higher in the second period than in the first is consistent with the conclusion of Gjerde et al. (2007), who also find that the Accounting Act of 1998 has contributed to increased value relevance of earnings.