We use the data published by the companies, not by data vendors or providers. By using the actual reports, we can be sure not only that we are using primary sources, but that any adjustments to previous years are identified and correctly entered into the database.
Where the jurisdiction requires tagged account submission, for example in XBRL, our team have sophisticated tools to help them overcome the problems that such tagging creates. Companies take great care to make sure that their printed or downloadable reports are accurate, and indeed spend large sums decorating them with photographs, cute type faces, charts and graphs. But they rarely have any interest in making sure that the XBRL / XML tagging required by regulatory authorities is done properly and well – it’s just a compliance overhead, to be dealt with as cheaply and quickly as possible. As a result, much XBRL data is actually of remarkably poor quality. We are expert in highlighting unlikely or poor tagging, and in sorting out the data so that the right numbers match the correct terms. (For example, US SEC filings contain over 120 different ways to tag “long term debt”, without including companies’ own unique, non-standard tags). Automated data extraction systems simply cannot match the accuracy we deploy through our dedicated staff.
To assess companies’ financial data usefully, the data needs to be strictly comparable between not just companies, but jurisdictions. Robur’s speciality lies in non-US companies, and we are set up to handle the different standards of translation. The translation of a Korean term to “Operating Income” might actually mean something very different to what a UK or Australian company means by that term. To be strictly comparable, the actual operating income might need to be identified under a different translated term, or calculated from several entries in the accounts.
Our team are expert at applying the correct term to the numbers. They understand company financials, because it is what they do all day, and they use templates and background checking software, purpose-written by us, to highlight improbable numbers or items which need checking. But above all, it is their brains which ensure that our standard terms are correctly matched to the appropriate data from the company statements, whether the company is a Chinese pressure-vessel manufacturer or a German medical equipment specialist.
When we process company reports, we don’t simply enter the latest numbers. The previous five years of data are automatically available, so that significant changes can be highlighted during the data encoding process, and numbers can be checked and verified. This is important also for ensuring that numbers which may not appear in the statements, but in the notes to the statements, such as diluted shares outstanding, or the “extraordinary items” which are used to calculate net income excluding extraordinaries, or the correct assignment of dividends to financial years, are correct and filled in.
Our team really care about accuracy. Nobody can truthfully claim to be 100% accurate, when handling millions of data items: mistakes can always be made. But our team do not make use of the phrase “It’s good enough”. They do their serious utmost to get it right, and they communicate with each other constantly, ensuring that they deploy a growing corpus of shared encoding expertise, and that anything unusual on the screens before them is handled in the most accurate and consistent way.