How to Generate Alpha From Hidden Earnings Data

Analysts who overlook the footnotes are missing investment opportunities.

Bigstock photo

Bigstock photo

While asset managers, allocators, and investment banks race to deploy machine learning and other artificial intelligence applications to help in the search for higher returns, critical information may be hidden in old fashioned footnotes. Of course, it may take a data scientist to unearth it.

To forecast earnings, analysts learn to isolate and identify the various components of a company’s net income. But a recent research paper published in the Journal of Financial Economics argues that even the best analysts miss a significant number of off-the-income-statement items that companies disclose — intentionally or otherwise — in less obvious parts of the 10-K, the annual comprehensive financial report that companies file with regulators. Identifying these items can give an analyst or investor a much clearer picture of the company’s actual earnings. The authors came to this conclusion by taking advantage of a novel database that allows users to more easily identify these hidden items.

“Core Earnings: New Data and Evidence,” a report authored by Professor Ethan Rouen and Eric So of the MIT Sloan School of Management and Professor Charles Wang of Harvard Business School, explains that the primary challenge for analysts is to quantify and distinguish core-earnings items — those that stem from companies’ recurring central activities — from non-core-earnings items, which are related to “ancillary business activities or transitory shocks.”

The larger backdrop for the professors’ analysis is the accounting practices at public companies. For years, companies have adjusted the financial figures they are required to report, particularly during crisis times like the pandemic, and try to focus investors on earnings that disregard costs or activities that they believe will be a one-time or rare event. While adjustments are allowed under accepted accounting standards, it makes analysts’ jobs harder as few have the time to go through details hidden in places other than the income statement.

Using proprietary data from the financial research firm New Constructs, the professors found that the average number of non-core earnings items disclosed in 10-Ks rose from six to eight between 1998 and 2017. They also found that about half of these non-core items were disclosed in footnotes, the cash-flow statement, or via management discussion and analysis, rather than in the actual income statement. This means that analysts or investors who are trying to understand the composition of GAAP earnings need to process a huge amount of information that’s often scattered in obscure parts of the 10-K, a task that some are unable or unwilling to undertake.

To make this daunting task easier, New Constructs mines data from footnotes of almost all publicly traded companies on U.S. exchanges and calculates an earnings distortion score for each company, which measures how much its core earnings deviate from its unadjusted net income. The Harvard and MIT professors found that about 19 percent of net-income items deviate from core-earning activities. They then constructed a core-earnings model that adjusts net income for items like net acquisition expenses, net currency expenses, and net legal expenses. They discovered that the adjusted net income could produce 8.2 percent annualized returns for investors who buy stocks that perform well on the earnings-distortion scale and sell those that perform the worst.

AltHub, a data integration platform that also used New Construct’s data to explore earnings distortion, came to similar conclusions. They found that once adjusted for earnings distortion, a portfolio tracking the S&P 500 could realize a 10-year annualized return of 13.9 percent, up from 12.1 percent for an unadjusted S&P 500 portfolio. “Professional investors who know about earnings distortion, or the degree to which companies manipulate earnings, can get an edge,” said David Trainer, CEO of New Constructs. He added that retail investors, meanwhile, can better protect their portfolios, while financial advisors can better protect and fulfill fiduciary duties for their clients. “Many companies exploit accounting rules to manipulate their earnings,” he said. “Investors should be putting in the time to examine the footnotes of earnings reports and financial statements to better understand a company’s risks.”