In recent years Securities and Exchange Commission (SEC) regulators have promoted the widespread and speedy dissemination of financial information to all users. The SEC Web site provides the following statement regarding its EDGAR archives of downloadable company financial reports and other filings (www.sec.gov/edgar/aboutedgar.htm):
“Its primary purpose is to increase the efficiency and fairness of the securities market for the benefit of investors, corporations, and the economy by accelerating the receipt, acceptance, dissemination, and analysis of time-sensitive corporate information filed with the agency.” [emphasis added]
In October 2000, the SEC Regulation FD (fair disclosure) became effective. Although it requires companies to disseminate important investor relations (IR) information via press releases, Regulation FD specifically encourages them to use the Internet to accomplish broad and rapid dissemination [1]. The Sarbanes-Oxley Act, signed into law in July 2002, requires larger companies to accelerate the filing of their Forms 10-Q and 10-K. In addition, these firms will be required to disclose in Form 10-K whether they provide these reports at their Web sites “as soon as reasonably practicable” after filing them with the SEC [8].
Corporate IR personnel also are vitally interested in using the Internet to rapidly disseminate IR information. The Standards of Practice for Investor Relations [10] states:
“Information should be released in a manner designed to reach the widest public audience possible, including the individual investor. Companies should encourage the use of multiple technologies to disseminate information.”
Corporate Web sites are especially suitable for distributing a wide variety of IR data, including analyst conference calls and manager presentations, since information can be posted in multiple formats (text, graphics, audio, and video) and languages.
Clearly, the rapid and widespread distribution of IR information via technology is desirable, both from the social (SEC) and corporate (investor relations) perspectives. Use of corporate Web sites for this purpose is an important activity with a relatively short history. Several studies have investigated determinants of the quantity and nature of financial data presented at Web sites (see [4, 5]). To the best of our knowledge, however, only one published study [6] has investigated the timeliness, as opposed to the extent, of online financial data. That study employs a sample of 47 firms, and tracks the time elapsed from the dates the firms’ Form 10-K financial reports are filed with the SEC until they appear on the firms’ Web sites. The study found an average lag of 30 days, with shorter lags associated with: greater corporate profitability, shorter lags in announcing earnings via press releases, and the use of multiple file formats for annual report/Form 10-K presentation. Longer lags are associated with external links to the SEC’s Web-based EDGAR archives.
The authors of [6] interpret their evidence as suggesting the following conclusions. Managers of more profitable firms have greater incentives to rapidly disseminate their firms’ financial results. Earnings release lags proxy for incentives to rapidly disseminate financial information not captured by other model variables. Use of multiple formats for 10-K/annual report files characterizes managers committed to providing high-quality IR materials, including rapid updating of financial reports online. Managers view external links to their firms’ 10-Ks at the EDGAR site as substitutes for providing the 10-Ks at their firms’ own Web sites. Finally, the authors of [6] argue that the capability to rapidly disseminate financial information online is irrelevant if managers do not have adequate incentives for rapid dissemination. They view the 30-day average update lag as indicating low incentives to update.
This study extends prior studies in a variety of ways. Prior research studied factors affecting the updating of one IR item at corporate Web sites: the 10-K/annual report. We investigate the updating of all IR items at corporate Web sites. The dependent variables are summarized in Panel A of Table 1. “Newest Content” equals the number of days between our Web site visit and the most recent dated material in a firm’s IR pages. The variable “Oldest Content” equals the number of days between our Web site visit and the oldest dated material in a firm’s IR pages. Stale material left at IR Web sites potentially creates legal liability [11]. A third variable, “Average Freshness,” was investigated, but models explaining this variable were not statistically significant.
Using Web crawler software allows us to dramatically increase sample size to 421 firms, versus 47 employed by [6]. The software automated data collection, capturing the structure and age profile of the IR portion of the firms’ Web sites.
This study adopts an innovative approach to investigating how large corporations manage their investor relations Web site. Rather than studying the impact of the investor relations Web site on the firm’s valuation, this study investigates the impact of various characteristics of the firm on the quality of its Web-based disclosures.
Research Sample and Procedures
Companies comprising the 2002 Fortune 500 list were used for this study. Financial information was obtained from Compustat. Using corporate IR addresses as starting points, Web information was collected using WebCat, a Web crawler program [7]. WebCat recursively searches and documents each site’s content and structure. Pages containing targeted keywords (see Figure 1) are dubbed “interesting,” which triggers the recursive search of all previously nonvisited hyperlinks found on that page. Failure to find any of these terms terminates the search along this branch of the tree.
This study focuses on that portion of the IR site under the direct control of the firm, as indicated by the Web page’s hosting domain. Each link is designated as Home-Based if the hosting domain name is associated with the targeted firm; otherwise it is designated as Outsourced.
File freshness is defined as the difference between the file download and upload dates reported in the file header. Some Web technologies made it impossible to determine the upload time of the Web page. After adjusting for time zone changes, 18% of the files were found to have an infeasible freshness value of negative one implying that the file was uploaded after it was downloaded in the study. The cause is likely a synchronization problem between the local computer used in the study and the remote-hosting servers. All lags were incremented by one for the analyses.
Data collection issues with the Web crawler software reduced the sample size to 421 from the original 500 firms. Three firms were eliminated due to mergers. Seventeen firms did not appear to have an IR Web site. The Web technology used on another 59 firms prevented the proper analysis of the site by the Web crawler software. Additionally, missing financial data for five firms further reduced the sample size to 416.
The explanatory variables, summarized in Panel B of Table 1, fall into two categories. First, we employ three variables to proxy for economic incentives prompting rapid IR updates (firm size, firm health, equity issuance). We anticipate that economic incentives are negatively associated with update lags.
Previous research indicates that larger firms are likely to provide more financial information, both in traditional media [3, 9] and online [5]. These findings suggest that the benefits of information dissemination increase with firm size, or that costs decrease with firm size, or both. Also, some researchers argue that larger firms are under greater pressure from analysts and investors to distribute their financial results quickly. For example, large firms typically release earnings announcements more rapidly than small ones [2]. Large firms also post their 10-Ks/annual reports more quickly at their Web sites [6]. Thus we expect larger firms will update their IR Web pages more rapidly. The association of firm size with both Newest and Oldest Content should be negative. “Firm Size” is measured as the natural logarithm of total assets, obtained from Compustat.
Firms having worse financial results to report, such as a lower return on sales, are slower to post their 10-Ks/annual reports at their Web sites [6]. We anticipate that firms in worse financial condition have more pressing IR problems than keeping their IR Web pages updated. We expect firms in poorer financial health will be slower to update their IR Web sites. The content of all IR Web pages is a much broader construct than is the provision of one 10-K/annual report. Therefore we employ as explanatory variable a broader measure of financial condition than the return on sales measure used in [6]. Our measure of “Financial Health” is the well-known Altman’s Z score (modified1), obtained from the Compustat database. The Z-score is a continuous measure associated with the firm’s likelihood of entering bankruptcy within the next year. Higher scores suggest better financial health, so we expect the association of Financial Health with both Newest and Oldest Content should be negative.
Firms that access the capital markets are more likely to engage in voluntary disclosure [9]. Managers have obvious incentives to disclose favorable information prior to a security offering. However, they also might disclose negative information if the market interprets not doing so even more unfavorably. Firms issuing new common stock equity in 1996 and 1997 disclosed more financial information at their Web sites in 1998 [5]. It is plausible that firms selling common stock will supply more timely information as well as a greater quantity. We expect equity issuance to be associated with more rapid updating of IR Web pages. Therefore, the association of an equity issuance variable with both Newest and Oldest Content should be negative. Our “Sell Stock” variable is coded “one” if a firm is a net issuer of common equity in 2000 and 2001, and “zero” otherwise.
Prior studies suggest that firms having high-quality IR efforts in one dimension tend to have high quality IR in other dimensions also. Ettredge et al. [5] found firms judged by analysts to have higher quality traditional IR disclosure practices also provide a greater quantity of IR materials at their Web sites. Ettredge et al. [6] observed a positive association between firms’ delays in releasing earnings announcements to the business press and their lags in posting 10-K/annual reports at their Web sites. The latter study also found that firms employing both PDF and HTML formats to present their 10-K/annual report information (rather than only one of the two), posted that report information to Web sites faster than other firms. They interpreted this result as indicating that firms more committed to high quality Web-based IR (as represented by use of multiple formats for documents) update their 10-K/annual report information more rapidly.
We extend those results by using four measures of a firm’s commitment to high-quality, Web-based dissemination of IR materials: number of IR pages, file format variety, percentage of bad links, and degree of local control. The metrics are intended to represent managers’ (unobserved) incentives to update IR data in timely fashion. The number of a firm’s IR pages, “N Pages,” provides a gross measure of site content. This variable should be negatively associated with age of content if it proxies for commitment to high-quality, Web-based IR. We capture variety and complexity of IR site content using “N File Types”: the number of different file extensions found at the site (such as .doc, .xls, and .pdf). Based on results in [6] we expect the association of this variable with content age to be negative. A variable equaling the percentage of bad links at the IR site, “Bad Home Links,” is an inverse proxy for commitment to quality. We expect the association of this variable with age of content should be positive. Finally, we measure commitment to quality by the extent of firms’ local control over IR pages, versus outsourcing. The “Local Control” variable equals the number of IR pages having a base URL associated with the firm divided by all IR pages visited. We expect the association of this variable with age of content should be negative.
Analysis and Results
Table 2 presents descriptive statistics for study variables. We are particularly interested in the distributions of the dependent variables. Both variables exhibit heavy clustering of observations having the values of “0” and “1.” For the Newest Content variable, at least three-fourths of the observations are coded “1” or less. For this reason, and after examining OLS regression residuals, we decided to employ ordinal logistic regression.
Table 3 presents the logistic regression results for dependent variables Newest Content and Oldest Content. Panel A contains results with Newest Content as the dependent variable. Three explanatory variables have significant coefficients. The sign of financial health is opposite to expectations. Although firms having higher current year profits might post their financial reports to their Web sites more rapidly [6], financially healthy firms do not provide the freshest general IR content at their sites. In fact, we find that the age of the newest IR content at firms’ sites is positively associated with financial health. A possible explanation is that less-healthy firms are diligent in publishing favorable information at their sites, although that information might not include the annual report (as in [6]). Such favorable information might include, for example, periodic management predictions of improved future performance, or frequent announcements of major customer orders.
The second significant variable, the number of file types provided at the IR site, is negatively associated with age of newest content, as expected. Our favored explanation is that number of file types is a proxy for commitment to high-quality Web-based IR, which implicitly includes fresh content. An alternative explanation is more prosaic: the richer the content delivery, the more likely it is that some of the content is very fresh. The third significant coefficient involves the percentage of bad home links at the IR site. If one views bad links as an inverse proxy for quality of Web-based IR, we would expect the percentage of bad home links to be positively associated with the age of newest content. Instead we observe the opposite. Fresher content is associated with a greater percentage of bad home links. We speculate that inserting new content sometimes disrupts links, and that firms identify such disruptions, and repair them, with a time lag. Thus, as the age of a firm’s newest content increases, the number of bad links tends to decrease.
Panel B of Table 3 indicates that age of oldest content is unaffected by firm size, issuance of stock, or percentage of bad links. Only one variable, N File Types, is significant in both panels, and the sign of that variable switches between panels. It appears that factors determining the age of oldest content are quite different from those affecting age of newest content. Stale IR content is interesting because it potentially exposes firms to liability risk at the hands of investors. The first significant coefficient in Panel B is number of IR Web pages. Our priors were that N Pages proxies for commitment to a high-quality, Web-based IR site. The result is opposite to this expectation. The age of oldest content is positively associated with N Pages. We propose that, as number of IR pages increase, so does the burden of keeping all information in those pages up to date. The second significant variable is N File Types. Again, our priors were that number of file types proxies for commitment to high-quality, Web-based IR, yielding an expected negative coefficient sign. Instead the estimated sign is positive. As with the number of IR pages, we speculate that, as richness of IR content increases, so does the likelihood that some of that content will be stale. Finally, the percentage of the IR content that is locally controlled rather than outsourced is positively associated with age of oldest content. This result, also, is opposite to the view that local control proxies for commitment to high-quality, Web-based IR. Apparently providers of outsourced IR content are more careful to prune stale content than are firms themselves.
Conclusion
The three firm-specific characteristics expected to affect timeliness of IR content—firm size, financial health, and sales of stock—proved to have limited influence. Only financial health is a significant variable, and its significance is limited to explaining the age of newest IR content. Firms having worse financial health tend to have fresher newest content, suggesting they use Web-based IR to deliver a message that better times lie ahead. The four proxies for quality of Web-based IR fare somewhat better in terms of significance: number of IR pages, number of file types, percentage of bad links within IR pages, and percentage of IR content hosted by the firm versus outsourced. Each of these variables is significant in at least one of the two regressions. However, the signs of these coefficients frequently differ from expectations, suggesting the variables do not always proxy for IR quality. As expected, firms having more IR file types have younger IR content. Richer content appears to proxy for a commitment to high quality IR including fresher content. However, newer IR content appears to be associated with more bad links at IR pages, suggesting that bad links sometimes are created when new content is inserted.
The variables that explain the age of newest content do not explain the age of oldest content. In particular, none of the proxies for high-quality Web-based IR have the expected signs. The estimated coefficients suggest the difficulty of pruning stale IR content increases as the number of IR pages and number of IR file types increase. In addition, firms that outsource more of their IR content have oldest content that is less old.
In summary, our results tell a complex story. We provide modest support, at best, for the view that high- quality IR content (as proxied by number of IR Web pages, number of IR file types, fewer bad links, and greater local control over content) is associated with fresher content. Differences between our results on this issue and results in [6] probably reflect differences in sample size (our sample is larger), and differences in the scope of IR content studied (our scope is broader). Our results have some fairly practical implications for Web-based IR and, by extension, for other Web-based corporate information. These include the following: After inserting new content, firms should check to determine whether any links have been disrupted. Firms posting extensive information at Web sites should be particularly careful to prune stale content. And firms that outsource part of their Web site content should determine whether old, self-hosted content is being pruned in parallel with outsourced content.
Join the Discussion (0)
Become a Member or Sign In to Post a Comment