پایداری سود، شرایط اقتصادی و ارتباط ارزش اطلاعات حسابداری

This study demonstrates that the value relevance of accounting information is influenced by the ability to capitalize investments in valuable resources. We use data from Sweden to show that firms that operate in industries in which accounting conservatism limits this capitalization display lower value relevance as a result of more unsustainable earnings components. However, when controlling for the different properties of sustainable and unsustainable earnings components, the difference vanishes. Moreover, we show that firms operating in industries in which more investments are immediately expensed display systematic temporal variations in the level of value relevance. We contend that economic conditions in the form of investment levels and growth expectations explain this variation. Thus, value relevance can be substantially affected by the prevailing economic context.

Accounting information plays an important role when shareholders evaluate a firm's prospects in forming their investment decisions. In accounting research, statistical associations between accounting information and share prices are used to assess the degree of value relevance of accounting information for shareholders (Collins et al., 1997, Gjerde et al., 2011 and Thinggaard and Damkier, 2008). Although a few studies report that the level of value relevance changes in the long term (Collins et al., 1997 and Francis and Schipper, 1999), there is little documentation of short-term variations in measures of value relevance. Temporal variations in value relevance are a matter of research interest because value relevance measures are often used to compare time periods and accounting regimes. Such comparisons rely on the assumption that measures of value relevance are solely determined by the accounting system. We use a sample of Swedish firms to challenge this important assumption and investigate how value relevance is associated with earnings sustainability and general economic conditions. The analysis departs from two realities: (i) an immediately expensed investment decreases current earnings but increases future earnings, and (ii) the unconditional form of accounting conservatism inhibits firms from capitalizing their investments in many valuable resources (e.g., research and human capital). We argue that because the level of investment varies over time, firms that invest heavily in resources that cannot be capitalized as assets display larger temporal variations in their reported earnings. The consequence is that the unconditional form of accounting conservatism has more severe effects on value relevance measures when a firm expenses more of its investments. The variation in the effect of a conservative accounting system over time is not an unknown feature (Givoly & Hayn, 2000), but its effects on measures of value relevance have attracted little attention in prior research. In the empirical analysis, we follow the procedures of researchers such as Francis and Schipper (1999) and partition the sample to study differences in value relevance between firms that are likely to capitalize a large portion of their investments in valuable resources (referred to as traditional industries) and firms that must expense most of their investments in valuable resources (referred to as non-traditional industries). The initial empirical tests suggest that firms operating in traditional industries, such as manufacturing firms, report more value-relevant information than firms in non-traditional industries, such as consulting and biotechnology firms. To better understand these differences in value relevance, we begin by analyzing reported earnings and their components. To be value-relevant, reported earnings must represent a level of earnings that can be sustained in the future (Beaver, Lambert, & Morse, 1980). We develop three models that separate sustainable from unsustainable earnings components. Throughout the study, we regard sustainable earnings as earnings components that are expected to prevail over a multi-period future, and we consider unsustainable earnings to be transitory, single-period earnings components. The empirical tests show that the separation of sustainable and unsustainable earnings components increases the measure of value relevance. However, value relevance increases considerably more for firms in non-traditional industries. Indeed, the difference in value relevance between the two groups disappears completely when sustainable and unsustainable earnings components are separated. Our results suggest that firms operating in non-traditional industries have more unsustainable earnings components in their reported earnings. Although the separation of sustainable and unsustainable earnings components is based on mechanical techniques rather than, for example, a firm's own disclosure of unsustainable earnings components, we believe that the larger unsustainable earnings components in non-traditional industries can be attributed to their many investments in valuable resources that cannot be capitalized. We contend that a comparison of the value relevance between two samples is complicated if the ability of reported earnings to capture sustainable earnings differs between the two samples. Next, we analyze whether economic conditions affect measures of value relevance. We suggest that the growth in GDP per capita proxies for the firm's level of investment, and that the stock market's average book-to-market ratio proxies for the market's growth expectations. We document the systematic temporal variations in value relevance for non-traditional industries and find that these variations are significantly associated with the level of investment and growth expectations. However, firms operating in traditional industries do not experience these variations. We suggest two explanations. First, an immediate expensing of valuable investments renders accounting earnings similar to cash flows, and when the level of investment is high, current earnings are particularly unrepresentative of future earnings. Thus, high investment levels reduce the value relevance of accounting information for firms with non-recognizable resources. When the level of investment is low, the reported earnings contain fewer unsustainable elements; thus, the value relevance increases. Second, firms that rely on unrecognized resources are more difficult to understand, and this difficulty reduces an investor's ability to determine their future cash flows and value. We suggest that the difficulty of understanding a firm's resource base plays a greater role when the expected growth rate is high than when it is low. Although the two proxies we use are somewhat crude measures of the investment level and growth expectations, the analysis clearly shows that exogenous factors influence the relationship between accounting information and value when investments are immediately expensed. Our results suggest that researchers must be cautious to avoid mistakenly attributing differences in value relevance to differences in accounting because the real cause can be the research design. We suggest two areas in which caution is required. First, biases are more likely to occur in comparisons between samples in which the proportion of unsustainable earnings elements differ. In these situations, simple adjustments for unsustainable earnings will provide a more reliable testing environment. Second, biases are more likely to occur in comparisons between samples containing differences in the level of investment and growth expectation. In particular, these biases can occur when the analyzed time periods are short. These analyses may include comparisons of the value relevance before and after the adoption of a new accounting standard. Indeed, when the firms in these samples have a greater reliance on resources that must be immediately expensed, more biases may be present in the analysis. The remainder of the paper is organized as follows. “Research hypotheses” discusses prior research and develops the hypotheses to be tested. “Measuring value relevance” outlines our research methodology and data sample. “Test results and analysis” presents the results of the empirical analysis, and “Conclusions” concludes the analysis.

Conventional conservative accounting standards hinder firms from capitalizing some of their investments in valuable resources; rather, these investments are immediately expensed. The consequence of these behaviors is that earnings become more akin to cash flows; thus, when substantial investments are undertaken, earnings decrease, even when such investments actually increase future performance. We show that firms undertaking more investments that are immediately expensed have more unsustainable earnings and lower value relevance. However, the effects of these unsustainable earnings components on value relevance are easily removed in our models. When we eliminate these effects, there is no difference in value relevance between traditional and non-traditional industries. Although the models that are used to decompose reported earnings into its sustainable and unsustainable components can remove differences in the level of value relevance, these models are incapable of removing differences in the temporal variation in value relevance. Firms operating in non-traditional industries, in which more valuable resources cannot be capitalized, consistently show more variation in value relevance. Moreover, these variations are systematic over time and associated with our two measures of economic conditions: the level of investment and growth expectations. When investment levels and growth expectations are high, firms operating in non-traditional industries display considerably lower value relevance. This finding is consistent, independent of the earnings decomposition model that is used. The empirical results are strong and exhibit a high level of statistical significance. However, because the interpretation of the results is rather open, a caveat is necessary. In particular, the reader should note that our empirical measures are not perfect proxies of our theoretical constructs; thus, there is ample space for future research. Our analysis relies heavily on the distinction between traditional and non-traditional industries. Although there are differences between these two industry groups, we do not know the specific causes of these differences. We believe that non-traditional industries undoubtedly have more non-capitalized resources; however, firms in non-traditional industries might also be younger and less diversified, may grow more rapidly, and may have different ownership structures. It is beyond the scope of this paper to examine the effects of these other factors on value relevance. If such differences exist and are able to explain the level of and temporal variation in value relevance, then such an investigation would be an interesting endeavor. A similar caveat exists for the test of value relevance under different investment levels. Because the investments that we investigate are typically undetectable in financial statements (i.e., they are immediately expensed and bundled with all other expenses), we use growth rates at the country level as our proxy. This measure could be correlated with other macro-economic variables. In fact, we have already shown that the change in GDP per capita and the book-to-market ratio are highly correlated. The patterns in the empirical study are relatively clear, but the interpretation is somewhat open. Again, further research is needed. One of the most important objectives of value relevance research is to study differences in the usefulness of various accounting methods from an investor perspective. We believe that it is of utmost importance that varying proportions of unsustainable earnings and changing economic conditions do not distort researchers’ conclusions regarding value relevance in different time periods and under different accounting regulations. For this reason, the models we use to decompose reported earnings can serve as a departure point in future comparative analyses. Moreover, the fact that the level of value relevance is systematically associated with measures of economic conditions suggests that future comparative analyses should control for differences in the economic conditions – particularly when firms in the sample are likely to immediately expense many of their investments.