The Mosaic of Corporate Communication: What Corporate Guidance Databases Leave Out, and Why It Matters
Every quarter, listed companies try to tell the market where they are heading. They forecast revenue, earnings, costs, capital spending and tax rates, among other financial items. They also talk about less easily quantified issues: the business environment, demand, inflation, industry conditions or the outlook for specific divisions. For investors, analysts and academics, this “management guidance” is one of the main ways companies communicate expectations about the future. But there is a catch. What firms actually say and what eventually appears in widely used financial databases are not necessarily the same thing.
That gap is at the heart of a new study by Xiaoxi Wu (Department of Accounting, Bocconi University), together with William J. Mayew of Duke University’s Fuqua School of Business and Jedson Pinto of the University of Texas at Dallas.
Their paper, On the usefulness of guidance reports, examines a vast and unusually detailed source of corporate forecasts: the Guidance Reports (GR) produced by the London Stock Exchange Group, or LSEG. These reports are compiled by specialists who review company press releases and transcripts of events such as earnings calls and industry conferences, extract forward-looking statements, classify them and preserve information about where they were made, how they were phrased and, in oral disclosures, who said them.
The study covers 23,738 Guidance Reports for 1,918 S&P 1,500 companies between 2005 and 2021, containing 1,735,241 individual guidance instances across 181 items.
Companies speak in words more often than databases suggest
Across the sample, 64.33% of all guidance instances are qualitative, compared with 35.67% that are quantitative. Companies may say that “demand is improving”, that an industry environment “remains difficult” or that revenue will be “relatively flat,” rather than giving a precise number. That matters because one of the most widely used research databases, I/B/E/S Guidance, focuses only on quantitative guidance for 13 items.
The difference is very significant. Of the 1.735 million guidance instances found in LSEG Guidance Reports, about 1.494 million do not appear in I/B/E/S Guidance for the same firm-years. This is not simply a technical difference between two databases. It changes the picture of how companies communicate.
Revenue is a good example. It is the most frequently guided financial-statement item in the sample, but much of that guidance is qualitative. Out of 200,574 revenue-guidance instances, 123,733 are textual rather than point or range estimates. The paper puts the conceptual point clearly:
“Since GR captures the verbatim text surrounding each guidance instance, scholars can use this feature to perform content or sentiment analysis.”
A forecast is not just a number. The words surrounding it can signal trends, confidence, uncertainty, caution or emphasis. They can also reveal why management believes a particular outcome is likely. The researchers find, for instance, that qualitative guidance tends to be accompanied by longer explanations and more positive language than quantitative guidance. Less precise forecasts generally come with more words. In practice, when managers cannot—or do not want to—compress the future into a number, they often compensate with narrative.
Guidance beyond earnings day
The study also challenges another simplified view of corporate forecasting: that guidance belongs mainly to earnings announcements.
Earnings calls and earnings releases are indeed crucial. In 99% of the firm-years studied, companies issued at least one guidance item around an earnings announcement. Managers provided an average of 16.71 distinct guided items in those periods. However, guidance also appears elsewhere: at broker conferences, industry events, analyst days, M&A updates and dedicated guidance calls. That makes the channel itself potentially meaningful. Managers may choose one venue for a carefully prepared statement and another for a more interactive discussion in which analysts can challenge assumptions.
Guidance Reports also identify the speaker. That makes it possible to distinguish, for example, what CEOs tend to discuss from what CFOs emphasize. In the sample, CFO-only guidance is especially common for income taxes, revenue, capital expenditure, EPS and operating profit, while CEOs are more strongly associated with business outlook and revenue.
This opens a different way of looking at voluntary disclosure: not merely “what did the firm forecast?”, but “who said it, where, and under what circumstances?”
The missing data are not by chance
The most consequential part of the paper concerns the gap between the richer Guidance Reports and I/B/E/S Guidance. The researchers show that the omissions are systematic.
Some are inevitable: qualitative guidance is excluded because I/B/E/S Guidance is numerical. Other items disappear because the database covers only a narrow set of metrics. But even when a quantitative guidance item could in principle be included, it is not always standardized and entered.
Suppose a manager says revenue will grow by 5%. To make that forecast directly comparable with an analyst consensus expressed in dollars, the data provider has to convert the percentage growth rate into a dollar figure. That requires additional information and additional work.
According to the study, when a Guidance Report forecast is expressed in a different unit from the corresponding I/B/E/S measure, its likelihood of being included falls by 32.4%. Alternative performance measures are also less likely to be included. Conversely, guidance is more likely to appear when managers repeat the same forecast or issue it around an earnings announcement, both signals that the information is likely to be important to users.
The authors summarize the implication in one of the paper’s most insightful sentences:
“Our findings suggest that the GR-IG discrepancy is not random.”
That is more than a warning about missing observations. Randomly missing data are one problem; systematically missing data are another. If certain kinds of forecasts are more likely to disappear—because they are harder to standardize, less followed by analysts or expressed in unconventional formats—then research based only on the standardized database may inadvertently study a selected version of corporate communication. The authors therefore caution researchers against assuming that readily available data represent the full population of management guidance.
A richer view of how companies talk about the future
The study does not simply argue that I/B/E/S Guidance is defective or obsolete. Quite the opposite: its standardized forecasts can be useful because they can be compared directly with analyst expectations, making it easier to measure forecast surprises, accuracy and bias.
The broader Guidance Reports can serve different purposes. They preserve the mosaic of real-world corporate communication: words as well as numbers, percentages as well as dollars, formal releases as well as conference conversations, and CEO statements as well as CFO explanations.
That difference may become increasingly important as financial research turns toward natural-language processing and artificial intelligence. A large archive of precisely identified forward-looking corporate statements gives researchers the chance to study not only whether a forecast was right, but how managers construct expectations in the first place.
