Nissim, D. 2024. Earning Quality, Fundamental Analysis and Equity Valuation, Columbia Business School. At https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3794378.

The following paper applies pattern recognition to detect anomalies in accounting data:

Pattern Recognition and Anomaly Detection in Bookkeeping Data, by Pierre Liang, Aluna Wang, Leman Akoglu, and Christos Faloutsos. Obtain the paper from Pierre Liang at the Tepper School of Business, Carnegie Mellon University.

Here’s a paper that applies large language models (LLMs) to identify sustainable earnings, testing alternative approaches. It leaves an open question: Can AI perform better than using business judgment to identify unsustainable items?

Shaffer, M. and C. Wang. 2024. Scaling Core Earnings Measurement with Large Language Models. At https://ssrn.com/abstract=4979504.

This paper applies machine learning to predict misstatements:

Bertomeu, J., E. Cheynel, E. Floyd, and W. Pan. 2021. Using Machine Learning to Detect Misstatements. Review of Accounting Studies 26 (2), 468-519.

Here is an example of the new presentation from the proposal:

Chapter 10 vividly reports the effect on net income when Delta Air Lines reversed its valuation allowance. Beware: If Boeing judges at any time in the future that it is more than 50% likely it will return to profitability, it will reverse the valuation allowance, increasing reported profit.  This is not sustainable profit. That is  bleed back.

Recognizing anticipated losses is good conservative accounting. However, there is something else to watch out for: If and when the planes are delivered and revenues recognized, cost of goods sold will be lower and profit margins higher because production costs have been written off. Those margins will not be because of reduced production costs but because of the earlier write-off. In 2020, although production proceeded with costs incurred, Boeing reported that production costs in inventory were near zero because of the write-off. So the cost of planes produced that year will have zero cost of goods sold when sold.

Up to 2002, firms would book an estimate of the total cost of a restructuring as a liability obligation then reduce the liability as costs were incurred. As explained in chapter 10, that led to the abuse of overestimating the liability than bleeding back any estimated cost not incurred to boost income in the future⸺the big bath problem. FASB Statement 146 and IAS 37 checked this practice by requiring restructuring charges to be booked mainly as they were incurred. That, however, had the effect of losing the information in the estimated liability for restructuring charges to be recognized over several periods in the future. And initial charges recognized as incurred might not be one-time items, leaving open the question of whether there will be further charges subsequently.

A case in point is Procter & Gamble (PG). By 2001, the firm had booked restructuring charges as they were incurred over seven quarters, totaling $1.3 billion, with indications the total would later go to $4.0 billion. With so many consumer products succeeding or failing, P&G’s restructuring charges are likely to be recurring. So are they really unsustainable items? As of 2024, the company was conducting an annual restructuring program “to maintain a competitive cost structure,” with annual restructuring charges ranging from $250 to $500 million. For the fiscal quarter ending September 2024, a restructuring charge of $886 million was booked. Better to see these charges as repeating costs of doing business?

By 2025, Xerox had reported impairments of goodwill in three out of the four previous years.

Amazon did the same in 2022, reducing depreciation expense by $3.6 billion and increasing net income by $2.8 billion. In early 2024, it said it would again reduce depreciation by further extending useful lives, increasing operating income before tax by $3.1 billion for the 2024 year.

See Laplante, Skaife, Swenson, and Wangerin, Journal of Accounting and Public Policy 38 (2019), 89-105.

That was partly explained by higher export growth. However, in the last quarter of 2024, Chinese firms were fast tracking their shipments to the U.S. to avoid the anticipated high tariffs on Chinese imports that the expected next U.S. president, Donald Trump was promising for 2025. That is channel stuffing. Look for a decrease in Chinese income from exports in 2025 as the inventory buildup from the channel studying reduces the demand for Chinese goods (as well as from lower exports because of the tariffs). That will have a negative effect on Chinese GDP growth for 2025, forecast to be 5% once more.

An example of sales generated by investment asset growth or not is same store sales versus sales from new store openings in retail. The P, Q, FX breakdown is discussed in chapter 10.

Beneish, M. 1999. The Detection of Earnings Management. Financial Analysts Journal 55 (5), 24-36.

The following sites give for more detail on the score and its implementation:

https://www.affaridiborsa.com/articoli/123-m-score-8-indici-in-uno-per-scovare-le-frodi-contabili.html

https://en.wikipedia.org/wiki/Beneish_M-score