«Leif Atle Beisland University of Agder Dissertation submitted to the Department of Accounting, Auditing and Law at the Norwegian School of Economics ...»
Several explanations for the loss of value relevance is offered in the existing literature, including an increased portion of non-information based stock trading (Dontoh, Ramakrishnan, and Ronen ), an increased volatility in stock returns (Francis and Schipper ), an increased frequency of negative earnings and non-recurring items (Collins, Maydew, and Weiss , Hayn ) and an increased change in pace (Lev and Zarowin ). These explanations are not mutually exclusive. Firms operating in an uncertain environment can be more prone to report losses and irregularities. Similarly, an accounting system’s failure to account for changes often creates losses and irregularities among firms.
Most studies dealing with long term changes in value relevance have to some extent discussed that firms increasingly rely on resources that have to remain unrecognized due to strict recognition requirements in conservative accounting systems (Collins, Maydew, and Weiss , Lev and Zarowin , Francis and Schipper , Goodwin and Ahmed ).
The consequences of this are that (1) the value of recognized net resources (equity) constitute a smaller portion of the firm’s market value of equity, and (2) investments in unrecognizable resources create more timing and matching problems in reported earnings. Collins, Maydew,
where many resources remain unrecognized. They find that the level of intangible-intensity has a negative association with value relevance. In contrast, Francis and Schipper  find no significant difference in the value relevance of earnings when comparing high-tech and low-tech stocks for the period 1952-1994. Although they find some evidence that balance sheet information explains a higher portion of the variability in prices for low-tech firms relative to high-tech firms, they document significant increases over time in the explained variability of this relation for both industry categories. They conclude that any evidence of a decline in value relevance cannot be attributed to the increasing number and importance of high-tech firms in the economy. We will develop these ideas further later in this section.
It is important to note that an increasing number of international studies find no decrease in value relevance. King and Langli  find no decrease in value relevance in Germany and the UK, but a decrease in Norway. However, in a more extensive and recent study, Gjerde, Knivsflå, and Sættem  re-examine Norwegian data and find no change in value relevance in the time period 1965 to 2004. Using Australian data, Goodwin and Ahmed  find no change in value relevance when controlling for losses. Taken together these studies suggest that the decrease in value relevance is primarily related to the United States.
We know of no study outside the U.S. that examines the long term value relevance in respect to different industries. Because of different results when using U.S. and international data, as well as the fact that nobody has used detailed industry data outside the U.S. it is difficult to form clear-cut a priori expectations. Our major concern in this study is the value relevance of firms operating in non-traditional industries relative to traditional industries. Regardless of if there is any time-series trend in the level of value relevance, we expect no difference in value relevance between traditional and non-traditional industries. We base this expectation on
the value relevance of financial information can be attributed solely, or even primarily, to the increasing number and importance of high-tech firms in the economy”.
Hypothesis 1: Accounting information is equally relevant in non-traditional and traditional industries.
In several past studies of the long-term value relevance of accounting information researchers have recognized the different properties of positive and negative earnings (e.g., Collins, Maydew and Weiss , Core, Guay and Van Buskirk , Francis, Schipper and Vincent , Goodwin and Ahmed ). The reason for doing so is that negative earnings simply represent an exception from a series of positive earnings that constitutes a (positive) value (see e.g. Hayn  or Collins, Pincus and Xie  for discussions of negative earnings and the value of equity). A firm for which losses are indicative of the future is simply a firm that has no value. Adjustments for the different properties of profits and losses tend to substantially improve the value relevance of accounting information. For example, Goodwin and Ahmed  report almost twice as high R2 when removing lossmaking firms.
But are losses the only non-recurring items that firms report? Clearly a conservative accounting system ensures that one-time items of a negative character are recognized earlier than those of a positive character (Basu , Watts , Ball and Shivakumar ).
Hence reported non-recurring losses are larger and more visible than non-recurring profits, but probably also less frequent. If non-recurring items are transitory and hence largely value
Both predictable and unpredictable changes in e.g. customer demand and input prices affect (positive and negative) earnings and make it vary over time. A conservative accounting system ensures that these known/predictable effects are accounted for after their occurence.
There is an abundant accounting literature on earnings persistence (see e.g., Penman , Ramakrishnan and Thomas , Penman and Zhang ) suggesting that the nature of events and accounting rules cause reported earnings to have several distinguishable components. To understand these it is vital to also understand firms’ long-term profitability prospects. Profitability is known to reverse to a firm and/or industry mean.2 Penman and Zhang  and  use the return on net operating assets to determine the mean reversion pattern in earnings. In a similar study, Fama and French  use the return on capital employed and find a negative autocorrelation over time. A most fundamental aspect of the investment strategy proposed by Sloan  is the finding that persistence (or mean reversion patterns) of cash flows and accruals differ. Most of these models do not take industry-differences into account (although at least Penman and Zhang  note that this probably would improve results further). Even simple models that just separate reported earnings into a sustainable and transitory component prove useful in investment strategies (e.g., Penman and Zhang , Anderson and Brooks ). We expect that adjustments for the different properties of sustainable and transitory components of earnings increase the overall value relevance of accounting information.
Hypothesis 2a: Measures of sustainable earnings improve the overall value relevance of accounting information.
There are also good reasons to believe that many managers choose to label non-recurring income increasing items as normal items. If investors are able to detect such earnings management these items are essentially irrelevant to investors.
A phenomenon documented by many including Freeman, Ohlson and Penman , Penman , Dechow , Sloan , Fama and French , Penman and Zhang  and .
vary systematically across industries. One effect of a conservative accounting system is that losses are recognized earlier than gains (e.g., Basu , Watts ). Bad news thus appear as a one-time transitory element whereas (normal) good news is spread over several future periods as gains are realized. We expect that firms’ investments are affected differently depending on the extent to which they are recognizable in the accounting system. The less able a firm is to capitalize its investments at the time of acquisition (because of accounting conservatism), the more affected by transitory elements its earnings is going to be. A growth in unrecognized investments reduces contemporary earnings and creates “hidden reserves” (c.f., Penman and Zhang ). Similarly, a reduction in the level of investment releases these reserves and increases contemporary earnings. We build on the arguments of Penman and Zhang  and suggest that there is a systematic variation in current earnings caused by unrecognizable investments. As firms operating in the non-traditional industries are more likely to make investments that have to be immediately expensed we expect these industries to benefit the most from the acknowledgement of different properties of sustainable and transitory earnings components. Hence the decomposition has a greater incremental effect on the value relevance as measured in non-traditional industries.
Hypothesis 2b: Measures of sustainable earnings improve the value relevance of the nontraditional industries relative to that of traditional industries.
According to the first hypothesis value relevance, measured as the long term mean explanatory power of accounting information on security prices/returns, is expected to be similar in traditional and non-traditional industries. A number of studies suggest that the
, Francis and Schipper , Gjerde, Knivsflå, and Sættem ) and it is obvious that variations around a mean can be industry dependent. To the best of our knowledge, no previous study has investigated differences in the long term inter-temporal variability in value relevance. Our aim is to shed light on the value relevance of accounting information as reported by firms operating in non-traditional industries relative to that of firms operating in traditional industries. We propose two different reasons as to why there are greater variations in the explanatory power of accounting information in non-traditional industries. First, as discussed extensively above, investments made by firms in non-traditional industries are to a greater extent expensed when acquired. Therefore their accounting earnings contain fewer accruals than what is found in traditional industries. An investment in a production facility is capitalized and depreciated over its expected economic life. However, a computer software developer or biotech firm has to expense most of its investment immediately although their investments are just as critical for future earnings as the production facility is for the manufacturing firm. The software developer and biotech firm are likely to produce accounting earnings that are more correlated with contemporary cash flows and less correlated with future cash flows, as compared with the manufacturing firm. For the non-traditional industries we expect that the lower ability to make use of accruals reduces the relative explanatory power of accounting earnings in times when the level of investment is high. The conservative accounting system requires a higher degree of verification for recognizing good news as opposed to bad news (Basu ) and therefore we expect that a faster recognition of bad news improves the value relevance for bad years within the non-traditional industries relative to the traditional industries.
affected by the economic conditions.
Firms operating in non-traditional industries have fewer recognized resources and hence a greater portion of their market value of equity is dependent on uncertain future cash flows, rather than verified values of current resources. Ceteris paribus, increased uncertainty generates more changes in investor expectations and hence also in share prices. In relation to this, Francis and Schipper  argue that an increased return volatility has a negative effect on the value relevance of accounting information. Similarly, Dontoh, Ramakrishnan, and Ronen  show that an increased amount of non-information based trading decreases the value relevance of accounting information. We expect that firms operating in non-traditional industries have fewer capitalized resources and investors have less certain information to anchor their expectations on. Noise traders are known to be more frequent in such trading environments, particularly in good times. We build on the ideas of Dontoh, Ramakrishnan, and Ronen  and suggest that the value relevance of accounting information is lower in the non-traditional industries when the stock market has a relatively high valuation. For example, in the IT-bubble of the late 1990s the values of information technology and telecommunication firms increased dramatically. There were few capitalized resources that contributed to this increase in value. On the other hand, when the bubble burst the values of these firms became closely aligned with fundamentals. We therefore suggest the opposite in bad times, namely that noise-trading decreases and hence that the relative value relevance of non-traditional industries increases.
Hypothesis 3b: The relative value relevance of non-traditional industries is negatively affected by market sentiments.
3.1 Research Design Value relevance is defined as the ability of financial statement information to capture and summarize information that affects firm value. Following a vast tradition in accounting research we study the value relevance by examining the statistical association between accounting information variables and the level of (change in) market value using multiple regressions. The explanatory power of a model (adjusted R2, hereafter R2) is therefore the primary measure of value relevance. A high explanatory power indicates that within a group of firms the market values are well reflected by the accounting information. We use a
traditional price regression for our first measure of value relevance: