do not necessarily reflect the views of UKDiss.com.
Value Relevance of Accounting Information for Intangible-Intensive Industries and the Impact of Scale: The US Evidence
Introduction
Many studies focus their interest on the value relevance of intangibles and non-intangibles industries, in the last couple of decades. One of the primary reason is because of the fast-growing technological sector, pushing more companies to invest in Research and Development (R&D). US General Accepted Accounting Principles (GAAP) requires that all of the expenditures for R&D and some other intangibles as patent, intellectual capital and other have to be expensed when they occurred and cannot be capitalized. Which often result in massive losses in the companies that invest heavily in intangibles, especially in their first years of their operations and making their financial statements less useful for investors.
In 1974, FASB enacted the Statement of Financial Accounting Standards (SFAS) No.2; its general rule requires expensing all of the R&D expenditure when they occur, with exceptions for the assets acquired for R&D activities that have an alternative future use. The assets, software that has alternative future use is capitalized, and its amortization is expensed.
The analysis is explicitly based on the value relevance of earnings and book value over twenty years from 1995 to 2015 for intangible and non-intangible industries and is replication of the study performed by Ciftci et al. (2014)
The study is important because with the fast-growing technology, many companies make a significant investment in research and development (R&D), and many believe that the current reporting in the financial statements are not useful for the investors. An observation confirmed by Amir and Lev (1996) about financial information for investors valuing pharmaceutical and technological companies that invest significantly in intangibles. However, Collins et al. (1997) note that there is not a difference between value relevance of accounting information for non-intangible intensive and intangible-intensive industries. In a similar research Francis and Schipper (1999) likewise did not find any significant difference in value relevance between high- and low-tech industries. An explanation for this difference according to Brown et al. (1999) is the lack of control for variation in scale that might bias the R2 from the regression of price to earnings and book value.
The purpose of this research is to bring light to the value relevance of accounting information for intangible intensive industries versus non-intangible intensive industries. The intangible intensive industries are the industries that heavily invest in R&D, like tech companies, drugs companies, telecommunications, etc.
The paper investigates the following research questions: First question is whether or not there is a difference in scale variability between intangible intensive industries and non-intangible intensive industries. To find an answer to this question, we compare the value relevance between intangible and non-intangible industries for a period of 20 years from 1995 to 2015 using data from CRSP and Compustat. Consistent with Ciftci et al. (2014) and Collins et al. (1997) our findings show that without controlling for the variability of the scale there is no distinction in the level of a temporal trend in the value relevance of earnings and book value between both industries.
The second question is whether there is a difference in the temporal trend in value relevance between both sectors. The findings point that the variability of scale is higher for intangibles industries than for non-intangible sectors for the period of the sample. When we control for variability of scale the value relevance of earnings and book value for intangible industries declined over time, while there is no significant decline for non-intangible industries. Our finding is consistent with Ciftci et al. (2014) and Collins et al. (1997), who also documented decrease (increase) in incremental value relevance of earnings (book values).
And the last question investigates whether and to what extent immediate expensing of intangible investments contribute to the decline in value relevance for intangible intensive industries. The results point that both means for earnings and book values are significantly greater under capitalization than under expensing of R&D for intangible industries, suggesting that intangible companies on average are not in study state.
This research is important because explain the differences between Amir and Lev (1996) and Collins et al. (1997) findings. With our examination, we show that the accounting information is less usable for intangible intensive companies, once we control for variability of scale. The second reason is that we expend the studies based on temporal trends in the value relevance of accounting information. Using the work of Brown et al. (1999) about the control of scale, we expend his research by dividing the sample into two industry groups. We also, extending the sample period of the study from 1995 to 2015. The last reason is that the findings of this study may have implications for the debate on the accounting treatment of intangibles. Our finding points that capitalization of R&D would increase the usefulness of accounting information for financial statement users. The study is relevant because, with the very fast-changing technological environment, many industries get involved in R&D or other types of intangibles and knowing that the accounting information is reliable and relevant will help investors to make correct decisions.
Literature Review
In today’s economy intangibles are an essential part of company’s success. The importance of intangibles and R&D particular in the last decades attracted many researchers to study the issues related to them. Previous studies point that the R&D expenditures are associated with future earnings, proposing that these expenditures have future benefits as other assets. However, GAAP does not treat R&D expenditure as an asset and require to be expensed when occurred. (FASB). Some researchers raise concern about the usefulness of accounting information due to the different treatment of R&D compare to the rest of the assets. Even though the majority of intangibles are immediately expensed, the benefits linked to them are realized much later and are not matched with previously expensed intangibles, which seriously disturbed the matching accounting principle (Lev and Zarowin, 1999).
Amir and Lev (1996) point that intangible assets contribute to the market value of the firm, even though they are not allowed to be recorded as assets under the current accounting rules. Furthermore, they conclude that information included in the financial statements is not useful for investors and other financial statement users when valuating companies with large amounts of intangibles. Amir and Lev (1996) discover that cash flows, earnings, and book values are irrelevant on a standalone basis when valuating cellular companies. They also state that accounting reporting system is “ill-equipped” to provide value-relevant information in fast developing tech-sector. However, they did not expend their research to the rest of the intangible intensive industries. In another study, Collins et al. (1997) record that the R2 of price to earnings and book value is somewhat higher for intangible intensive than non-intangible intensive industries. Likewise, Francis and Schipper (1999) find that high-tech companies have similar value relevance compare to low-tech companies. Thus, the evidence provided by Amir and Lev (1996) is in conflict with the results recorded by Collins et al. (1997) and Francis and Schipper (1999).
In a similar study, Lev and Zarowin (1999) explained the difference in results because the accounting information on average is not distorted for the intangible intensive sector. They state that the earnings will be almost the same under R&D expenses capitalization or when the expensed immediately in steady R&D environment. However, they did not elucidate if R&D expense is in a steady state or whether the accounting information is distorted on average for intangible intensive industries.
Another study that investigates the value relevance of accounting information for an intangible intensive sector is one performed by Monahan (2005). In his research, he explores whether the adjusted accounting numbers for R&D expensing improved the value relevance. He observes that the capitalization of R&D expense statistically significant increase in value relevance when the companies have high future R&D growth. However, in his study, he does not take into consideration of coefficient of variation of scale factor when compare intangible with non-intangible intensive sectors. He also doesn’t examine the difference either in an intertemporal pattern in value relevance between both groups.
The research conducted by Brown et al. (1999) examine the impact of scale factor on value relevance through adding a variable for the scale factor. He further analytically shows that there is a positive association between R2 and coefficient of variation of the scale factor, and omitting the scale factor will bias not only the coefficient estimate but also the R2 of the regression. One of the reasons that affect the scale factor is the stocks split repurchases, differences in ROE and payout ratios. Thus, they suggest when comparing different groups based on R2 we should control for scale, that can be done by control variation in scale is regressing R2 from the price regression on a coefficient of variance of price and book value.
Collins et al. (1997) point that the value relevance of historical cost financial reports declines because of the move from an industrialized to highly technological economy. They offer couple reasons for this intertemporal drop in value relevance of accounting information. One of them is that the intangible intensive firms increased over time, taking into consideration the result from Amir and Lev (1996) study that value relevance of accounting information is less value relevant, the intertemporal increase in the intangible intensive firm should lead to the decline in value relevance of accounting information. Another reason is the growing firms that reporting loss, Hayn (195) indicates that negative earnings could lead to a loss in value relevance of accounting information. Based on this, Collins et al. (1997) study the value relevance of earnings and book value over forty years. They found out that the value relevance of book value and earnings increase instead to decline, but the bottom line earnings have declined, while the book value increases. Francis and Shipper confirmed these results.
However, Brown et al. (1999) argue that the intertemporal increase in value relevance is due to scale effect which acts as a correlated ignored variable. Brown et al. (1999) use coefficient of variation of book value and share price as a proxy for CV. Their results show that once we control for scale factor, the value relevance of book value and earnings decline. They also point that the increase of the value relevance in prior studies is due to the increase of coefficient of variation of the scale factor. Still, Brown et al. (1999) did not study the value relevance for intangible intensive and non-intangible intensive industries individually.
Core et al. (2003) record a substantial decline in value relevance of accounting information in the period of four years between 1995 to 1999, which they call “new economy” period (NEP) compares to previous years. A recent study executed by Dichev and Tang (2008) show that there have a mismatching of revenue and expenses, and this mismatch leads to a decline in persistence and predictability of earnings.
Core et al. (2003) record a decline in value relevance of accounting information in the period of four years between 1995 to 1999, which they call “new economy” sub-period compares to previous years. A recent study executed by Dichev and Tang (2008) show that there have a mismatching of revenue and expenses, and this mismatch leads to a decline in persistence and predictability of earnings.
Lastly, many studies explore the value relevance of R&D as one of the most essential types of intangible investments (Chan et al., 2001; Ciftci and Cready, 2001; Lev and Zarowin, 1999) argue that the capitalization of R&D not matter if company’s R&D is in steady state. Consequently, one probable explanation could be that the capitalization of R&D for both intangible and non-intangible industries will not lead to a change in the accounting numbers, thus will be in steady state, adopting the argument of Amir and Lev (1996). However, if the R&D is growing over time as some of the prior studies suggest (Lev and Zarowin, 1999; Ciftci and Cready, 2001), the reasonable explanation will be that the earnings and book value under R&D could be significantly higher under capitalization than those under expensing.
Hypothesis development:
Many studies focus their interest on the value relevance for intangible and non-intangible industries in the last couple decades. Amir and Lev (1996) find that the accounting information for intangible industries is not value-relevant, while Collins et al. (1997) argue that the combined R2 for intangible-intensive industries for book value and price is higher than that for non-intangible. In another study, Francis and Shipper (1999) find that high tech companies have value relevance similar to low tech industries. The results of these examinations are very inconsistent; therefore, the first hypothesis is:
Hypothesis 1: After controlling for the variability of scale the value relevance of book value and earnings is significantly lower for intangible intensive industries than non-intangible intensive industries.
Next, we investigate the temporal trend of value relevance and variability of scale. The expectations are that if the intangible intensive industries had a greater variability of scale as Brown et al. (1999) recorded, an increase in intangibles over time will lead to increase in variability of scale. Thus, the expected result in this study is that the temporal increase in variability of scale for intangible intensive sector will be higher.
Hypothesis 2: Intangible intensive industries have more substantial temporal increase in variability of scale.
If intangibles decrease the value relevance of accounting information the increase in intangibles will lead to bigger temporal decline in value relevance for intangible industries.
Hypothesis 3: Intangible industries have a greater temporal decline in value relevance than for non-intangible industries
Next, we check the temporal decline after 2000, that is associated with technological boom. The hypothesis is as follow:
Hypothesis 4: In the years after 2000, the temporal decline in value relevance of accounting information reverses for intangible intensive industries
Lastly, we investigate how the capitalization of R&D will affects the value relevance of earnings and book value. The expected result is that the value relevance will increase significantly with the capitalization of R&D instead of expensing it.
Hypothesis 5: Capitalization of R&D will increase the value relevance for intangible intensive industries.
Methods:
Following Ciftci et al. (2014) and Collins et al. (1997) we use R2 from the regression of book value and earnings as a measure of value relevance. We use Ohlson model (1995) consistently with Collins et al. and Ciftci et al., and estimate the price regression individually for each group: intangible intensive and non-intangible intensive industry:
Pit = 0 + 1EPSit + 2BVPSit + it (1)
Pit = 0 + 1EPSit + it (2)
Pit = 0 + 1BVPSit + it (3)
Where:
Pit – is the price per share for firm i three months after fiscal year end in year t, adjusted for stock split using adjustment factor from CRSP;
EPSit – is the earning per share for firm i in fiscal year t, and is calculated as Net Earnings divided by Shares outstanding;
BVPSit – is the book value per share, calculated as book value of equity divided by the shares outstanding; and
it – is other value-relevant information. All of the above variables are taken from Compustat.
The first equation examines the value relevance of combined earnings and book value while the second and third examine the relationship between EPS and price, and BVPS and Price per share. We use equations (2) and (3) to calculate R2 from BVPS and EPS alone regressions. In all of the equations, the P is the dependent variable, and BVPS and EPS are independent. We do not expect any economically significant difference between both groups without controlling for CV of a scale factor, consistent with Ciftci et al. (2014) and Collins et al. (1997).
Then examine whether there is any difference in the temporal trend in CV of scale factor between intangibles and non-intangibles industries. (Hypothesis: 2) For this examination, we use the approach used by Brown et al. (1999) where BVPS and Price per share are used as a proxy for the scale factor. Then we estimate the following regressions to evaluate the temporal trend in the CV of scale for both intangible and non-intangible industries separately.
CVgt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + it (4)
Where:
CVgt – is the coefficient of variation for group g in year t, its either CV_BVPSgt or CV_Pgt.; CV_BVPSgt – is the coefficient of variation of book value for group g in year t, is calculated as standard deviation of the BV divided by its mean; CV_Pgt – is the coefficient of variation of price per share for group g in year t, is calculated as standard deviation of the price per share divided by its mean; INT_D – is a dummy variable that equals 1 for the intangible intensive industry and 0 for the non-intangible intensive industry, and shows the difference in intercept between intangible and non-intangible industries; TIMEt – is calendar year t minus 1995, and shows the temporal trend in CV of scale for both groups.
Hypothesis 2 forecast that the CV of scale for intangible intensive industries is increasing faster than for non-intangible industries, which means that the coefficient estimates for INT_D*TIMEt should be positive. The sum of INT_D*TIMEt and TIMEt indicates the trend for intangible intensive industries and the F test of significance level of the sum is
F-test-1 = INT_D*TIMEt + TIMEt = 0
To test hypothesis 2 and 4, we explore the impact of CV’s of scale on Value relevance and the time trend in Value – relevance using Ciftci et al. (2014) and Brown et al. (1999) model.
R2gt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + 4CV_Pgt + 5CV_BVPSt + it(5)
R2gt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + 4D2000t + 5D2000t*TIMEt +6D2000t*TIMEt*INT_D +7CV_Pgt + 8CV_BVPSgt + it(6)
Where:
R2 is generated from equation (1) for group g in year t for either intangible and non-intangible industries; D2000t – is dummy variable that equals 1 for years greater than 1999, 0 otherwise.
To test hypothesis 1 we perform a F-test 2: [INT_D + 12.5(INT_D*TIME) = )], where 12.5 is half of the 25 years period for the research).
The Hypothesis 3 predicts a temporal decline in value relevance of accounting information is greater for intangible than non-intangible intensive industries, so the expectation is the coefficient estimate for INT_D*TIMEt to be significant and negative.
Equitation (6) is to test Hypothesis 4; and the hypothesis 4 predict that the value relevance will reverse in the years after 2000, compare to the prior period for intangible industries, or that
D2000t*TIMEt +D2000t*TIMEt*INT_D will be positive and significant and that is F-test-3 to test H4.
To investigate the temporal trend in the incremental value-relevance of earnings and book value as per Collins et al. (1997) and Ciftci et al. (2014) using the R2 from equations (1) to (3)
INCR_R2Et = R2Tt – R2Bt,
INCR_R2Bt = R2Tt – R2Et,
Where R2Bt/Et – is the R2 of book value/earnings alone in year t from equation (2) and (3); R2Tt – is the total R2 from equation (1); and INCR_R2Et/Bt is the incremental R2 for book value /earnings for year t.
Lastly to test Hypothesis 5, or how R&D capitalization impacts the value relevance for intangible industries we use the following equation:
Pit = 0 + 1EPSAJit + 2BVPSAJit + it (1’)
Pit = 0 + 1EPSAJit + it (2’)
Pit = 0 + 1BVPSAJit + it (3’)
Where:
EPSAJit – is the earning per share under R&D capitalization, and is calculated as EPS + RD expenses – amortization expense; BVPSAJit – is the book value per share under R&D capitalization, and is calculated as BKVL + R&D expenses – amortization expense; The price per share is the dependent variable, and BVPSAJ and EPSAJ are independent variables.
To calculate the amortization, we use straight-line depreciation with five years useful life as in Ciftci et al. (2014), Chan et al. (2001).
Hypothesis 5 forecasts that the Value-relevance of accounting information under capitalization will be greater than under expensing. Thus, R2 for equation (1’) should be greater than from equation (1), also should have a significant difference between equation (2) and (2’) and (3) and (3’). To control for the impact of variability of scale on the R2 from book values and earnings we use the following model:
R2gt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + 4CV_Pgt + 5CV_BVPSAJt + it (4’)
Where: R2gt – is the R2 from equation (1’) for group g in year t; and BVPSAjit – is the book value per share under R&D capitalization. If capitalization increases value relevance, the differences in R2 for intangible and non-intangible industries should reduce.
Sample Selection:
The research uses a sample selection that consists data obtained from Compustat and the Center for Research and Security Prices (CRSP) databases. The Price per share and market value of equity is taken from CRSP, while the rest of the variable are from Compustat. The sample includes observations from the period of 1995 to 2015 for all publically traded companies listed in NYSE (New York Stock Exchange). Since the focus of the paper is to compare intangible with non-intangible industries the first year of an investigation will be 1995, so we can capture any difference before 2000 and after 2000, or when the technology boom takes place.
The sample period of the research performed by Cifti et al. (2014), the paper that has been replicated is from 1975 to 2006. In his study, Ciftci et al. (2010) pick 1975 as first year because the SFAS No. 2 was put into inception, while Collins et al. (1997) research is for forty consequent years, and focus in investigating the intertemporal pattern in value relevance over four decades.
Results:
As mention above, the sample includes firm-year observation between 1995 and 2015. Consistent with the Ciftci et al. (2014) and Collins et al. (1997) all of the observation for the book value of equity and total assets greater than zero has been deleted. Also, all of the observation at top and bottom 1.5% of earnings, book value to market value are removed, as well as 1.5% of one-time items, and all of the values of Price per share equal or greater than $1,000.
The initial sample contained more than 200,000 firm-year observations, after removing all of the outliers we left with 29,311 observations, where 9,413 are from intangible intensive industries and 19,899 form non-intangible intensive industries. In this research, we use the definition of intangible-intensive industries same as Collins et al. (1997).
The first table shows the descriptive statistics. From the results on table 1, panel A is clear that the average intangible intensive industries are less profitable than non-intangible industries. Consistent with Ciftci et al. (2014), both earnings (E) and book value (BV) of the intangible intensive group are considerably smaller, while the average price per share for intangible industries is smaller with a very small percentage. The market price (MV) for both industries are similar, which suggest that the difference in earnings and book value might be a result of the accounting treatment of intangibles.
Compatible with Collins et al. (1997) and Ciftci et al. (2014), the mean of RDS (R&D expenses to sale ratio) is 14.146 for intangibles is significantly higher for intangible industries than non-intangible industries where is 0.981. Also, market value (MV), ONE for intangible industries is higher than non-intangible industries, as the frequencies of loss (LOSS). Collins et al. (1997) propose that the higher frequencies of losses may reduce value relevance, even though they didn’t study the difference between intangible and non-intangible industries. Panel B shows that the frequencies of loss and one-time items increase over time, but the increase is greater for intangible industries.
TABLE 1:
Panel A: Descriptive Statistics a
| Mean | Median | |||||
| Variables: | INT | Non-INT | P-value of difference | INT | Non-INT | P-Value of Difference |
| P | 21.486 | 22.444 | 0.00 | 10.64 | 14.93 | |
| EPS | -0.996 | 0.417 | 0.00 | -0.077 | 0.524 | |
| EPSAJ | 89.877 | 76.30 | 0.00 | -2.552 | 2.870 | |
| BVPS | 5.862 | 9.333 | 0.00 | 3.495 | 6.868 | |
| BVPSAJ | 96.735 | 85.217 | 0.00 | 17.843 | 13.641 | |
| RDS | 14.146 | 0.981 | 0.08 | 0.135 | 0.012 | |
| ONE | -0.298 | -0.324 | 0.00 | 0.000 | 0.0009 | |
| LOSS | -35.444 | -30.042 | 0.00 | -1.478 | 0.00 | |
| ROE | 2.278 | 16.048 | 0.401 | -0.557 | 2.072 | |
| DIVP | 0.006 | 0.017 | 0.00 | 0.000 | 0.000 | |
| CV_EPS | -1.564 | -2.136 | 0.00 | 10.526 | 0.811 | |
| MV | 3,370,632.53 | 3,350,130.98 | 0.00 | 262.96 | 539.739 | |
| N | 9413 | 19,899 | ||||
Panel B: The frequency of loss and one-time items
| Loss Frequency LOSS | One time percentage ONEP | |||
| INT | Non-INT | INT | Non-INT | |
| 1995-2000 | ||||
| 2001-2005 | ||||
| 2006-2010 | ||||
| 2010-2015 | ||||
a The number of firm-year observation from Compustat and CRSP is 29,311, after cleaning the data of all negative book values, outliers, and share prices less than a dollar
b Variables: P- is the price per share of a firm three months after year end in year t adjusted for stock splits. EPS – is the earning per share in year t, calculated as Net Income (NI) divided by numbers of shares outstanding (CSHO). EPSAJ –is EPS under capitalization of R&D expenditures, calculated by adding R&D expense per share (XRD-from Compustat) and deducting hypothetical amortization expense per share. Hypothetical amortization expense is calculating by amortizing R&D assuming five years useful life. BVPS- is the book value per share in year t, calculated as book value of equity (CEQ) divided by the number of shares outstanding. BVSAJ is BVPS under capitalization of R&D expenditures, calculated by adding the R&D capital to BV of equity. RDS is R&D expense to sale ratio, ONE is one-time items per share, calculated as sum of extraordinary items (XI), discontinue operations (DO), and special items (SPI), divided by the number of share outstanding. Loss is frequency of losses; the firm is considering to be in loss when its core earnings are less than zero. Core earnings are calculated as EPS minus ONE. ONEP is percentage of one-time items calculated as absolute value of ONE as a percentage of core earnings. ROE is return of equity, calculated as NI divided book value of equity. DIVP is the dividend payout ratio, calculated as dividends (DVC) divided by market value of equity. CV_EPS – is coefficient of variation of EPS, is calculated as standard deviation divided by its mean. INT (intangible intensive industries) are SIC codes: 282 – plastic and synthetic materials;283 drugs; 357 computer and office equipment; 367 – electronic components and accessories; 48- communications; 73 – business services; and 87 – engineering. Non-INT are non-intangible intensive industries.
c p-Value of T-statistics for the difference between intangible and non-intangible industries (two-sided)
The next step in this paper is to investigate the differences in factors that affect the variability of scale. Brown et al. (1999) suggest that increases in numbers of shares reduce scale, so the differences in dividend payout ratios and performance (ROE) between both industries should lead to a difference in scale. The results also show the difference in mean and median coefficient of variation of EPS (CV_EPS), proposing that cross-sectional variation in earnings are more substantial for intangible industries. The overall result implies that there are significant differences between intangible and non-intangibles sectors, as per Brown et al. (1999) suggestion that affects the variability of scale.
In order to examine whether the intangible industries are in steady state we observe the impact on R&D capitalization on book value and earnings. For intangible industries, the mean and median for EPS is -0.996 and -0.077 respectfully, while the mean and median for EPSAJ is 89.87 and -2.55, showing very significant difference. The mean and median for BVPS is as follow: the mean is 5.862 and median 3.495 and for BVPSAJ is 96.735 and 17.843 respectfully, showing also significant difference. Taken together the result suggest that on average book value and earnings are lower under expenditure of R&D than under capitalization for intangible intensive industries, which point that intangible industries are not in steady state.
Table 2
Value relevance of accounting information for intangible and non-intangible industries without controlling for variability
| Model: Pit = 0 + 1EPSit + 2BVPSit + it (1)
Pit = 0 + 1EPSit + it (2) Pit = 0 + 1BVPSit + it (3) |
|||||||||
| Equitation (1)
(A) |
Equitation (2)
(B) |
Equitation (3)
(C) |
(A-C)
INCR_Ea |
(A-B)
INCR_Ba |
|||||
| 1 | 2 | Adj. R2 | 1 | Adj. R2 | 1 | Adj. R2 | |||
| Non-INT | -2.167
(-30.15) |
2.518
(51.154) |
0.324 | -3.041
(-38.546) |
0.136 | 2.871
(57.349) |
0.259 | 0.065 | 0.188 |
| INT | -0.231
(-3.898) |
1.807
(86.992) |
0.278 | 0.5
(7.248) |
0.003 | 1.795
(87.289) |
0.277 | 0.001 | 0.274 |
a The number of firm-year observation from Compustat and CRSP is 29,311.
b Intangible intensive and non-intangible intensive industries and variables are in Table 1.
c This table present the mean coefficient from cross-sectional estimation of equation (1) to (3).
The result of t-statistics is present in parentheses
INCR_E is incremental R2 from EPS generated as R2 from (1) minus (3). INCR_B – is incremental R2 from BV from (1) minus (2).
Table 2, presents the R2’s from the price regression of equation (1) to (3). The results in the first column are from equitation (1) for both industries, R2 for non-intangible industries is 0.324 and for intangible is 0.278 from outcomes we can see that there is not significant difference between both. The outcome is consistent with Ciftci et al. (2014) and Collins et al. (1997) even though the period in this research paper is different. The second column represent the R2 from equation (2), where we include in the regression only earnings. The R2s for non-intangible industries is 13.6% while for intangible is only 0.3%, which shows that the difference between them is significant at 5% level. The next column is for equation (3) and the results for R2 for non-intangible and intangible industries are as follow: 25.9% and 27.7% the difference again is not statistically significant, even though the R2 for intangible is slightly higher. The last two columns represent the incremental R2 from EPS and BV. From the table 2, we can see that the incremental information provided from earnings is higher for non-intangible industries compare to intangible, while the incremental information provided from BV is higher for intangible intensive industries. These results suggest that value relevance for BV is higher most likely because the expensing is less severe for BV than EPS.
Table 3, bellow presents the mean values of the coefficient of variance for CV_P, CV_BV and CV_EPS for four sub-periods for both sectors. The results show that the coefficient of variation for all of the variables is higher for intangibles industries compare to non-intangibles industries, even for CV_EPS which we don’t use as proxy for scale and is added only for demonstration purposes. The results also point that the increase in CV is over time.
Panel B of table 3 shows the Pearson correlation between some of the variables used in equation (6). The results display that CV_P and TIME are highly correlated and positive as well as the relationship between CV_BVPS and TIME. The INT_D is also positive and highly correlated with the proxies, suggesting that intangible industries have higher coefficient of variances.
TABLE 3
PANEL A: DESCRIPTIVE STATISTICS a
| Mean | |||
| Variables c: | SUB PERIOD | INT | NON-INT |
| CV_P | 1995-2000 | 1.371 | 1.272 |
| 2001-2005 | 1.261 | 1.162 | |
| 2006-2010 | 1.214 | 1.143 | |
| 2010-2015 | 1.204 | 1.111 | |
| CV_BVPS | 1995-2000 | 1.096 | 1.042 |
| 2001-2005 | 1.046 | 0.984 | |
| 2006-2010 | 1.037 | 0.962 | |
| 2010-2015 | 1.010 | .0.952 | |
| CV_EPS | 1995-2000 | -0.975 | -1.764 |
| 2001-2005 | -1.564 | -2.136 | |
| 2006-2010 | -1.376 | -2.454 | |
| 2010-2015 | -1.233 | -3.133 | |
PANEL B: PEARSON CORRELATION:
| TIME | INT_D | CV_P | CV_BVPS | |
| CV_P | 0.066** | 0.063** | ||
| CV_BVPS | 0.079** | 0.077** | 0.642** | |
| CV_EPS | 0.001 | 0.001 | 0.002 | -0.006 |
a The number of firm-year observation from Compustat and CRSP is 29,311.
b Intangible intensive and non-intangible intensive industries and variables are in Table 1.
Panel A presents the mean value of CV’s for sub periods. Panel B presents the Person correlations for the whole sample period.
Variables c: TIMEt – year t -1995; INT- is and dummy variable that equal 1 for intangible intensive industries, and 0 for non-intangible intensive industries; CV_P –is the coefficient of variation of share price in each year, calculated as the standard deviation of share price divided by its mean; CV_BVPS – is the coefficient of variation of book value per share in each year, calculated as the standard deviation of share price divided by its mean; CV_EPS – is the coefficient of variation of earnings per share in each year, calculated as the standard deviation of share price divided by its mean;
**Correlation is significant at the 0.01 level (2-tailed).
Table 4 present the temporal trend in the coefficient of variance of scale for CV_P and CV_BVPS from equations (4). The results suggest that the CV_P and CV_BVPS are positive and significant related to TIME. Also, INT_D*TIME is significant to CV_BVPS indicates that the increase in coefficient for variance is higher for intangible industries then non-intangible industries. The result in table 3 overall support Hypothesis 2.
TABLE 4. TEMPORAL TREND IN CV_P AND CV_P
| CV_P | CV_BVPS | |
| TIME | 0.009 (9.591) *** | 0.006 (12.302) *** |
| INT_D | 0.137 (5.794) *** | 0.064 (5.75) ** |
| INT_D*TIME | 0.002 (1.293) | 0.050 (2.138) ** |
| R2 | 0.096 | 0.116 |
| aF-test-1 | 14.206*** | 17.279*** |
| aF-test-2 | 15.414*** | 19.077*** |
The table presents the coefficient estimates from Equation (4)
F-test-1 test whether (TIMEt +TIME*INT_Dt) = 0
F-test-2 –whether [TIMEt +(10*TIME*INT_Dt)] = 0
CVgt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + it (4)
Variables: TIMEt – is yaer t minus 1995; INT_D – is a dummy variable that equals 1 for INT and 0 for non-INT industries. CVgt – is either CV_Pgt or CV_BVPSgt. CV_Pgt and CV_BVPSgt Are defined in Table 1.
*Statistically significant at 10%
**Statistically significant at 5%
***Statistically significant at 1%
Table 5, below present the outcome form equation (5) and (6). The first part, Panel A presents the outcome when the dependable variable is R2 from regressing price on BVPS and EPS for the intangible and non-intangible intensive sectors. The results show that the TIME is not significant, which mean that here is not a temporal trend for non-intangible intensive industries. The next column shows the result when we controlling for coefficients of variance proxies, from the results we can see that the TIME itself is not significant but the TIME*INT_D is highly significant. Thus, Hypothesis 3 is accepted. The results are very similar for the last two columns, showing that when we control for VC of proxies the TIME*INT_D became significant. And that support Hypothesis 4, suggesting that the temporal decline reverses at the endo of the NEP. The other two panels respectfully show the results for the incremental R2 for EPS and BVPS. The incremental R2 for EPS shows that the TIME is highly significant when we control for CVs of proxies, and also when we considering D2000 and D2000*TIME.
TABLE 5 TEMPORAL PATERN IN TOTAL AND INCREMENTAL VALUE RELEVANCE
PANEL A:
| EQUATION (5) | EQUATION (6) | |||
| INT | NON-INT | INT | NON-INT | |
| TIME | -0.004
(-0.340) |
-0.003
(-0.109) |
-0.002
(-0.097) |
-0.007
(-1.991) *** |
| INT_D | 0.0098
(0.173) |
-0.005
(-0.23) |
0.012
(0.042) |
-0.003
(-0.01) |
| INT_D*TIME | -0.0047
(-0.983) |
-0.007
(-2.377) *** |
-0.079
(-1.566) |
-0.008
(-2.131) *** |
| D2000 | -0.321
(-4.162) *** |
-0.322
(-2.361) |
||
| D2000*TIME | -1.203
(-3.152) *** |
-0.872
(-2.144) *** |
||
| D2000*TIME*INT_D | 0.085
(0.91) |
0.006
(0.78) |
||
| CV_P |
|
0.000
(0.224) |
0.840
(0.202) |
|
| CV_BVPS | -0.003
(-2.151) *** |
0.002
(1.418) |
||
| R2 | 0.083 | 1.09 | 0.284 | 0.270 |
PANEL B: INCREMENATAL R2 FROM EPS
| EQUATION (5) | EQUATION (6) | |||
| INT | NON-INT | INT | NON-INT | |
| TIME | -0.007
(-0.425) |
0.009
(0.492) |
-0.025
(-7.01) *** |
-0.031
(-14.815) *** |
| INT_D | -0.012
(-0.73) |
-0.032
(-0.127) |
-0.023
(-0.984) |
-0.006
(-1.03) |
| INT_D*TIME | -0.004
(-0.325) |
0.004
(0.983) |
0.000
(-0.003) |
-0.013
(-1.436) |
| D2000 | -0.076
(-2.52) *** |
-0.037
(-3.613) *** |
||
| D2000*TIME | -1.203
(-3.152) *** |
-0.030
(-6.523) *** |
||
| D2000*TIME*INT_D | 0.007
(0.931) |
0.013
(0.431) |
||
| CV_P |
|
-0.095
(-1.63) |
-0.001
(-1.113) |
|
| CV_BVPS | -0.018
(-1.506) |
-0.005
(-2.158) |
||
| R2 | 0.240 | 0.18 | 0.205 | 0.264 |
PANEL C INCREMENTAL R2 FROM BVPS
| EQUATION (5) | EQUATION (6) | |||
| INT | NON-INT | INT | NON-INT | |
| TIME | 0.012
(0.737) |
0.012
(0.709) |
0.127
(3.755) *** |
0.129
(0.810) |
| INT_D | -0.053
(-0.974) |
-0.074
(-1.542) |
-0.034
(-0.717) |
-0.003
(-0.973) |
| INT_D*TIME | 0.013
(2.034) |
0.008
(1.892) |
0.006
(1.325) |
0.004
(0.371) |
| D2000 | 0.489
(2.061) *** |
0.474
(2.206) |
||
| D2000*TIME | -0.143
(-3.011) |
-0.872
(-2.144) *** |
||
| D2000*TIME*INT_D | 0.140
(1.375) |
0.035
(1.262) |
||
| CV_P |
|
0.541
(2.315) *** |
0.249
(1.154) |
|
| CV_BVPS | -0.004
(-0.541) |
0.077
(1.767) |
||
| R2 | 0.542 | 0.24 | 0.711 | 0.728 |
The table presents the coefficient estimates from Equation (5) and (6)
T-statistics are reported in parenthesis.
INT and Non-INT industries are defined in Table 1, as well as all of the variables. Panel A shows the coefficient estimate when the dependent variable is total R2 from equation (1). Panel B – shows R2 from equation (2) when the dependable variable is EPS. Panel C – R2 from equation (3) when the dependent variable is BVPS.
R2gt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + 4CV_Pgt + 5CV_BVPSt + it (5)
R2gt = 0 + 1TIMEit + 2INT_D + 3INT_D*TIMEt + 4D2000t + 5D2000t*TIMEt +6D2000t*TIMEt*INT_D +7CV_Pgt + 8CV_BVPSgt + it(6)
Variables: TIMEt – is year t minus 1995; INT_D – is a dummy variable that equals 1 for INT and 0 for non-INT industries. D2000 is a proxy variable that equals 1 if year is greater than 1999 and 0 otherwise. CV_Pgt or CV_BVPSgt are defined in Table 1.
*Statistically significant at 10%
**Statistically significant at 5%
***Statistically significant at 1%
The last table 6, present the outcome when we capitalized R&D expenditure. The first panel represent the regression table for equation (1’), (2’) and (3’) or when the R&D expenditure are capitalized. Panel B of the table represent the differences between adjusted R2 for earnings and book value between table 2, and Panel A of table 6. The result point that the capitalization of R&D increases the total R2 for intangible intensive industries near 25%, while the increase for non-intangible intensive industries is less than 2%. The difference amongst both groups is significant. In general, the results in table 6, backing up Hypothesis 5.
TABLE 6. THE IMPACT OF CAPITALIZATION OF R&D ON VALUE RELEVANCE OF EARNINGS AND BOOK VALUES
PANEL A
| Model: Pit = 0 + 1EPSAJit + 2BVPSAJit + it (1’)
Pit = 0 + 1EPSAJit + it (2’) Pit = 0 + 1BVPSAJit + it (3’) |
|||||||||
| Equitation (1’)
(A) |
Equitation (2’)
(B) |
Equitation (3’)
(C) |
(A’-C’)
INCR_Ea |
(A’-B’)
INCR_Ba |
|||||
| 1 | 2 | Adj. R2 | 1 | Adj. R2 | 1 | Adj. R2 | |||
| Non-INT | -2.372
(-66.23) |
2.379
(66.57) |
0.326 | 0.011
(-8.961) |
0.008 | 0.013
(10.55) |
0.012 | 0.314 | 0.318 |
| INT | -0.168
(-80.59) |
1.682
(80.95) |
0.254 | 0.007
(12.504) |
0.008 | 0.008
(14.181) |
0.10 | 0.154 | 0.246 |
PANEL B
| Model: Pit = 0 + 1EPSit + 2BVPSit + it (1) Pit = 0 + 1EPSAJit + 2BVPSAJit + it (1’)
Pit = 0 + 1EPSit + it (2) Pit = 0 + 1EPSAJit + it (2’) Pit = 0 + 1BVPSit + it (3) Pit = 0 + 1BVPSAJit + it (3’) |
|||||||||
| Equitation
(1)-(1’) (A) |
Equitation (2)-(2’)
(B) |
Equitation (3)-(3’)
(C) |
(A-C)-
(A’-C’) INCR_Ea |
(A-B)-
(A’-B’) INCR_Ba |
|||||
| 1 | 2 | Adj. R2 | 1 | Adj. R2 | 1 | Adj. R2 | |||
| Non-INT | 1.93 | 0.139 | 0.002 | -3.031 | 0.128 | 2.858 | 0.247 | 0.249 | 0.13 |
| INT | -0.063 | 0.125 | 0.023 | 0.493 | 0.005 | 1.787 | 0.177 | 0.154 | 0.028 |
The table presents the main R2 from cross- sectional estimation of equation (1), (2) and (3) under expensing the R&D and under capitalization of R&D (1’), (2’) and (3’) and the differences between them.
INT and Non-INT industries are defined in Table 1, as well as all of the variables.
Variables: INT_D – is a dummy variable that equals 1 for INT and 0 for non-INT industries. EPS, BVPS, EPSAJ and BVPSAJ are defined in Table 1.
Conclusion:
This paper investigates the value relevance for intangible intensive industries and non-intangible intensive industries for US companies listed in NYSE. First we extend the period from 1995 to 2015. Further, we investigated the impact of variability of scale on the value relevance and found out that once the control for variability for scale, the value relevance is significantly lower than that for non-intangible intensive industries and there is a temporal decline for the intangible sector. We also investigate the difference in value relevance once we capitalized the R&D expenditure. The results show that once the R&D is capitalized the overall value relevance increase for intangible industries, but did not completely eliminated, which suggest that there are more factors that affect the value relevance.
The limitation of this studies are as follow: we did not examine the different scale related effects of capital market; we also did not took into consideration the impact of the IFRS, for multinational companies.
References:
Amir, E., and B. Lev. 1996. Value-relevance of nonfinancial information: The wireless communications industry. Journal of Accounting and Economics, 22, 3-30.
Brown, S., K. Lo, and T Lys. 1999. Use of R2 in accounting research: measuring changes in value relevance over the last four decades. Journal of Accounting and Economics, 28, 83-115.
Chan, L., J. Lakonishok, and T. Sougiannis. 2001. The stock market valuation of research and
development expenditures. Journal of Finance, 56, 2,431-2,457.
Ciftci, M and Cready, 2001
Ciftci et al 2014
Collins, D., and S. Kothari. 1989. An analysis of intertemporal and cross-sectional determinants of earnings-response-coefficients. Journal of Accounting & Economics, 11, 143-181.
Collins, D., E. Maydew, and I. Weiss. 1997. Changes-in the value-relevance of earnings and book values over the past forty years. Journal of Accounting and Economics, 24, 39-67.
Core, J., W. Guay, and A. Buskirk. 2003. Market valuations in the New Economy: an investigation of what has changed. Journal of Accounting and Economics, 34, 43-67.
Dichev, I., and V. Tang. 2008. Matching and the changing properties of accounting earnings over the last 40 years. Accounting Review, 83, 1460-1425.
FASB, 1974, Statement of Financial Accounting Standards (SFAS) No.2
Francis, J. and K. Schipper. 1999. Have financial statements lost their relevance? Journal of Accounting Research, 37, 319-352.
Hayn, C. 1995. The information content of losses. Journal of Accounting and Economics, 20, 125-153.
Lev, B. and P. Zarowin. 1999. The boundaries of financial reporting and how to extend them. Journal of Accounting Research, 37, 353-385.
Monahan S. 2005. Conservatism, growth and the role of accounting numbers in the fundamental analysis process. Review of Accounting Studies, 10, 227-260.
Ohlson
