An Exploratory Study of the Effect on Forecast Accuracy of Using Different Geographic Segment Data Sources

Abstract

The specific research question this study investigates is whether or not model-based earnings forecasts utilizing COMPUSTAT's country specific geographic segment data produce more accurate earnings forecasts than model-based forecasts utilizing data from actual geographic segment footnotes. This study compares the accuracy of earnings forecasts from models using the geographic segment information in actual geographic segment footnotes to the accuracy of earnings forecasts from models using the geographic segment information in the COMPUSTAT business segment tape. Evidence collected in this study indicates that geographic segment data provided in the COMPUSTAT business segment data base appears to make it possible to more accurately predict sales than does the geographic segment data provided in the same companies' actual geographic segment footnote.Accounting research often involves the use of computerized data bases such as those compiled by the Center for Research in Security Prices (CRSP) and Investors Management Sciences (COMPUSTAT). Because of their importance to accounting research, these data bases have themselves been closely examined. For example, studies that have relied on computerized data bases have been criticized (McElreath and Wiggens, 1984) because it has been shown that these types of data bases (i.e., CRSP and COMPUSTAT) may contain incorrect data and inconsistent use of definitions. The purpose of this study is also to investigate accounting data bases, but with a focus on the enhanced geographic segment information on the COMPUSTAT business segment data base. Specifically, this study compares the accuracy of earnings and sales forecasts from models using the information in actual geographic segment footnotes to the accuracy of earnings forecasts from similar models that instead use the enhanced geographic segment information in the COMPUSTAT business segment tape.

The materiality of country-specific geographic segment disclosures

“…3 Only Canada and the U.K. were disclosed by a sufficient number of firms to be incorporated in tests of association with security prices (Boatsman, Behn and Patz, 1993;Nichols, Tunnell and Waldrup, 1996;Thomas, 2000). Using a simulation approach to assess the predictive ability of geographic segment information, Herrmann (1996) is able to expand this group of countries to include France, Germany, and Japan.…”

Are Foreign Earnings Disclosures Value-Relevant? Disaggregation Solves the Puzzle

“…As a consequence, segmented foreign earnings disclosures constitute relevant news for investors. Authors unanimously agree that the geographical disaggregation of a firm's performance conveys additional informative news to investors and analysts (Hussain and Skerratt, 1992;Herrmann, 1996;Nichols et al, 1996;Doupnik and Seese, 2001;Kou and Hussain, 2007). However, given the complex set of issues associated with this task, there is no doubt that there are numerous reasons why systematic mispricing errors may arise when investors assess the relationship between a firm's foreign activities and its shareholder value.…”

“…While most studies consider the decomposition of earnings into domestic versus foreign components (Bodnar and Weintrop, 1997;Thomas, 2000a;Christophe, 2002;Khurana et al, 2003;Bodnar et al, 2003a;Thomas, 2004;Callen et al, 2005;Hope and Kang, 2005), only a few authors have investigated how far investors incorporate the pricing effects of disclosed foreign operations performance into specific geographic regions (Boatsman et al, 1993;Thomas, 2000b;Christophe and Pfeiffer, 2002). However, the use of too aggregated foreign performance measurements mixes foreign earnings disclosures together, which originate from a range of distinctive markets whose stability and growth opportunities may be extremely heterogeneous and differentially perceived by investors (Hussain and Skerratt, 1992;Herrmann, 1996;Nichols et al, 1996;Doupnik and Seese, 2001;Kou and Hussain, 2007). The aggregation of these disparate effects leads inevitably to a difficult interpretation of the information -and hence to a downward bias in the estimation of the value-relevance of these aggregated This study adds to the existing literature by testing and comparing how markets assess the value-relevance of aggregated foreign versus domestic earnings and, more importantly, how investors value region-specific versus country-specific foreign performance disclosures.…”