Identifying Inconsistencies in Company Financials

Over the past few years, workflow automation has rapidly increased bringing along new efficiencies for asset managers. However, manual processes are still prevalent when analyzing company financials throughout the investment lifecycle, adding risk which can materially affect investment or trading decisions – especially when it comes to private markets, and credit managers who are dealing with complex asset classes like syndicated loans, private credit and CLOs. 

An average credit analyst might have 20-30 names they are responsible for. That means each quarter they need to read over 50 documents. The analyst will look at the usual items like balance sheet, income statement and cash flow, but will also look at marketing material and compliance certificates; they’ll need to deal with unaudited vs audited financials and restated financials, credit agreements, and unexpected events like an acquisition or refinancing, where a company’s capital structure can change significantly. There is no standard way to represent this information and there are often multiple sources and versions of an important metric. This information is used to value a company, make investment decisions, and conduct ongoing risk management of the portfolio.

Often, there are varying interpretations of what the best “metric” is to value a company. One of the hardest (and most important) is EBITDA. Most portfolio managers or sponsors don’t use GAAP EBITDA to measure earnings. Instead, an industry convention uses Adjusted EBITDA, which adds back certain expenses to give a more level view of a credit depending on the industry being evaluated. But the methodology could vary. The same can be said about where it’s best to source debt. For example, not all documents will show you details like the lien structure or subordination of bonds and loans. And as with any human activity (even with the best analysts), mistakes can be made that lead to bad decisions that  an impact on an organization. This is because every analyst will perform their role slightly differently, can simply get things wrong or have a different interpretation of a company’s financial information. If you were to ask three different analysts to compute earnings per share of a public company, you might get different answers depending on what they consider earnings. Add in the manual overlay and this can get very error-prone digging through documents, cross-checking information, and, in some cases, financial information may be recorded incorrectly.

To add more complexity to the equation, a company’s financial data can often be fragmented. This makes it challenging to access all the information they need while feeding the data into critical functions like portfolio and CLO compliance. It’s also often the case that a firm’s front and back office functions are different, requiring disparate systems to get the data (and needing to access financial documents from different locations). So what can be done to solve these issues?

A Front-to-Back Solution

People want as close as they can get to a single solution with a front-to-back ability to handle all of the functions needed to run a fund. They want order management, portfolio monitoring, asset servicing, investment research, pricing, analytics and risk all in one place. That’s why at Siepe, we knew that bringing in automated financial data extraction was crucial to closing the circle to a singular data platform, and providing a unique capability that few people have.

An issuer’s documents are electronically consumed into the platform and run through a series of steps that categorize the contents into relevant financial types and structures such as debt statements, cash flow statements, income statements and EBITDA. The system then analyzes the text, tables and other identifiers which describe a company’s performance and builds patterns to link these structures together as the issuer files new documents each month. With this, we can paint a picture from different perspectives – by looking through at the various financial KPIs, labels and statements, at different periods, or across the life of the issuer. 

While some portion of what you see in one of these documents is more or less standard and repeatable, the majority is not. It’s surprising how much complexity and change can occur between two reporting periods that should naturally be the same thing. It’s almost unpredictable the set of permutations, which makes this problem hard to solve.. Using machine learning, we build up a history of these patterns and we can use these for consuming new material and look for new patterns across industries, and improve our accuracy as time goes on. 

As an example, our platform has detected cases where a company inadvertently understated the amount of debt on its balance sheet. If not caught, or if you don’t know where to look for the “right” value (in one case it was after an acquisition), it could be problematic calculating leverage where it could be significantly higher or lower, making an investment less or more attractive than it actually is. We remove the error-prone nature of how people analyze these documents, allowing managers to come to a conclusion faster with the confidence of putting in a winning trade quicker than people who don’t.

Standard and Fully Customizable 

When we pull a company’s financials to we sort them into two different views to help mitigate the risk of any discrepancies:

  1. Standard default view
    This is a normalized view of all these different companies’ variations in reporting financial positions into a standard taxonomy that allows us to accumulate more data onto our platform. As a result, we can compare companies apples-to-apples from a financial performance perspective and sort them all into a standard taxonomy.
  2. Account specific
    If we’re working with a particular client, they may want to see certain figures, ratios or calculations, and not be interested in other things. Some investment managers will have their own issuer summary pages – like a dashboard or performance report for all of the most relevant market data, financial, PnL and other referential data about an issuer they have or are looking at. That’s why we have developed a flexible methodology for creating templates which are essentially a manager’s own tear sheet or issuer page, that can be made up of any data from the Financials Platform but could also be combined with other analytic data from our platform – all within one system. We work with them to determine the right data and calculations to employ. If they want to see leverage metrics or revenue in a certain light, all of that information is customizable through our front-end.

To find out more about how our investment management platform, data and financial extraction solutions can help you, contact us today.